Browse Source

fix doc (#3631)

* fix doc

* fix en doc

* fix link in doc

* fix img link
liuhongen1234567 8 months ago
parent
commit
349f983dee
100 changed files with 5064 additions and 1583 deletions
  1. 58 15
      docs/module_usage/tutorials/cv_modules/3d_bev_detection.en.md
  2. 58 17
      docs/module_usage/tutorials/cv_modules/3d_bev_detection.md
  3. 55 15
      docs/module_usage/tutorials/cv_modules/anomaly_detection.en.md
  4. 49 15
      docs/module_usage/tutorials/cv_modules/anomaly_detection.md
  5. 56 15
      docs/module_usage/tutorials/cv_modules/face_detection.en.md
  6. 50 15
      docs/module_usage/tutorials/cv_modules/face_detection.md
  7. 56 15
      docs/module_usage/tutorials/cv_modules/face_feature.en.md
  8. 49 15
      docs/module_usage/tutorials/cv_modules/face_feature.md
  9. 56 15
      docs/module_usage/tutorials/cv_modules/human_detection.en.md
  10. 49 15
      docs/module_usage/tutorials/cv_modules/human_detection.md
  11. 56 15
      docs/module_usage/tutorials/cv_modules/human_keypoint_detection.en.md
  12. 56 15
      docs/module_usage/tutorials/cv_modules/human_keypoint_detection.md
  13. 54 13
      docs/module_usage/tutorials/cv_modules/image_classification.en.md
  14. 47 13
      docs/module_usage/tutorials/cv_modules/image_classification.md
  15. 56 15
      docs/module_usage/tutorials/cv_modules/image_feature.en.md
  16. 49 15
      docs/module_usage/tutorials/cv_modules/image_feature.md
  17. 56 15
      docs/module_usage/tutorials/cv_modules/image_multilabel_classification.en.md
  18. 49 15
      docs/module_usage/tutorials/cv_modules/image_multilabel_classification.md
  19. 56 15
      docs/module_usage/tutorials/cv_modules/instance_segmentation.en.md
  20. 56 15
      docs/module_usage/tutorials/cv_modules/instance_segmentation.md
  21. 56 15
      docs/module_usage/tutorials/cv_modules/mainbody_detection.en.md
  22. 49 15
      docs/module_usage/tutorials/cv_modules/mainbody_detection.md
  23. 56 15
      docs/module_usage/tutorials/cv_modules/object_detection.en.md
  24. 49 15
      docs/module_usage/tutorials/cv_modules/object_detection.md
  25. 54 13
      docs/module_usage/tutorials/cv_modules/open_vocabulary_detection.en.md
  26. 54 13
      docs/module_usage/tutorials/cv_modules/open_vocabulary_detection.md
  27. 53 12
      docs/module_usage/tutorials/cv_modules/open_vocabulary_segmentation.en.md
  28. 53 12
      docs/module_usage/tutorials/cv_modules/open_vocabulary_segmentation.md
  29. 56 15
      docs/module_usage/tutorials/cv_modules/pedestrian_attribute_recognition.en.md
  30. 49 15
      docs/module_usage/tutorials/cv_modules/pedestrian_attribute_recognition.md
  31. 57 15
      docs/module_usage/tutorials/cv_modules/rotated_object_detection.en.md
  32. 56 15
      docs/module_usage/tutorials/cv_modules/rotated_object_detection.md
  33. 56 15
      docs/module_usage/tutorials/cv_modules/semantic_segmentation.en.md
  34. 49 15
      docs/module_usage/tutorials/cv_modules/semantic_segmentation.md
  35. 56 15
      docs/module_usage/tutorials/cv_modules/small_object_detection.en.md
  36. 49 15
      docs/module_usage/tutorials/cv_modules/small_object_detection.md
  37. 56 15
      docs/module_usage/tutorials/cv_modules/vehicle_attribute_recognition.en.md
  38. 49 15
      docs/module_usage/tutorials/cv_modules/vehicle_attribute_recognition.md
  39. 56 15
      docs/module_usage/tutorials/cv_modules/vehicle_detection.en.md
  40. 49 15
      docs/module_usage/tutorials/cv_modules/vehicle_detection.md
  41. 56 15
      docs/module_usage/tutorials/ocr_modules/doc_img_orientation_classification.en.md
  42. 49 15
      docs/module_usage/tutorials/ocr_modules/doc_img_orientation_classification.md
  43. 49 15
      docs/module_usage/tutorials/ocr_modules/formula_recognition.en.md
  44. 49 16
      docs/module_usage/tutorials/ocr_modules/formula_recognition.md
  45. 65 21
      docs/module_usage/tutorials/ocr_modules/layout_detection.en.md
  46. 59 22
      docs/module_usage/tutorials/ocr_modules/layout_detection.md
  47. 56 15
      docs/module_usage/tutorials/ocr_modules/seal_text_detection.en.md
  48. 49 16
      docs/module_usage/tutorials/ocr_modules/seal_text_detection.md
  49. 56 15
      docs/module_usage/tutorials/ocr_modules/table_cells_detection.en.md
  50. 56 15
      docs/module_usage/tutorials/ocr_modules/table_cells_detection.md
  51. 56 15
      docs/module_usage/tutorials/ocr_modules/table_classification.en.md
  52. 56 15
      docs/module_usage/tutorials/ocr_modules/table_classification.md
  53. 56 15
      docs/module_usage/tutorials/ocr_modules/table_structure_recognition.en.md
  54. 49 15
      docs/module_usage/tutorials/ocr_modules/table_structure_recognition.md
  55. 56 15
      docs/module_usage/tutorials/ocr_modules/text_detection.en.md
  56. 49 15
      docs/module_usage/tutorials/ocr_modules/text_detection.md
  57. 55 13
      docs/module_usage/tutorials/ocr_modules/text_image_unwarping.en.md
  58. 49 15
      docs/module_usage/tutorials/ocr_modules/text_image_unwarping.md
  59. 57 20
      docs/module_usage/tutorials/ocr_modules/text_recognition.en.md
  60. 67 20
      docs/module_usage/tutorials/ocr_modules/text_recognition.md
  61. 56 15
      docs/module_usage/tutorials/ocr_modules/textline_orientation_classification.en.md
  62. 57 15
      docs/module_usage/tutorials/ocr_modules/textline_orientation_classification.md
  63. 14 0
      docs/module_usage/tutorials/speech_modules/multilingual_speech_recognition.en.md
  64. 14 0
      docs/module_usage/tutorials/speech_modules/multilingual_speech_recognition.md
  65. 57 16
      docs/module_usage/tutorials/time_series_modules/time_series_anomaly_detection.en.md
  66. 49 15
      docs/module_usage/tutorials/time_series_modules/time_series_anomaly_detection.md
  67. 56 15
      docs/module_usage/tutorials/time_series_modules/time_series_classification.en.md
  68. 49 15
      docs/module_usage/tutorials/time_series_modules/time_series_classification.md
  69. 56 15
      docs/module_usage/tutorials/time_series_modules/time_series_forecasting.en.md
  70. 49 15
      docs/module_usage/tutorials/time_series_modules/time_series_forecasting.md
  71. 56 15
      docs/module_usage/tutorials/video_modules/video_classification.en.md
  72. 56 15
      docs/module_usage/tutorials/video_modules/video_classification.md
  73. 56 15
      docs/module_usage/tutorials/video_modules/video_detection.en.md
  74. 56 15
      docs/module_usage/tutorials/video_modules/video_detection.md
  75. 44 17
      docs/pipeline_usage/tutorials/cv_pipelines/3d_bev_detection.en.md
  76. 44 17
      docs/pipeline_usage/tutorials/cv_pipelines/3d_bev_detection.md
  77. 48 18
      docs/pipeline_usage/tutorials/cv_pipelines/face_recognition.en.md
  78. 49 18
      docs/pipeline_usage/tutorials/cv_pipelines/face_recognition.md
  79. 48 18
      docs/pipeline_usage/tutorials/cv_pipelines/general_image_recognition.en.md
  80. 49 18
      docs/pipeline_usage/tutorials/cv_pipelines/general_image_recognition.md
  81. 48 18
      docs/pipeline_usage/tutorials/cv_pipelines/human_keypoint_detection.en.md
  82. 24 12
      docs/pipeline_usage/tutorials/cv_pipelines/human_keypoint_detection.md
  83. 43 16
      docs/pipeline_usage/tutorials/cv_pipelines/image_anomaly_detection.en.md
  84. 46 16
      docs/pipeline_usage/tutorials/cv_pipelines/image_anomaly_detection.md
  85. 43 16
      docs/pipeline_usage/tutorials/cv_pipelines/image_classification.en.md
  86. 46 16
      docs/pipeline_usage/tutorials/cv_pipelines/image_classification.md
  87. 43 16
      docs/pipeline_usage/tutorials/cv_pipelines/image_multi_label_classification.en.md
  88. 46 16
      docs/pipeline_usage/tutorials/cv_pipelines/image_multi_label_classification.md
  89. 43 16
      docs/pipeline_usage/tutorials/cv_pipelines/instance_segmentation.en.md
  90. 46 16
      docs/pipeline_usage/tutorials/cv_pipelines/instance_segmentation.md
  91. 43 16
      docs/pipeline_usage/tutorials/cv_pipelines/object_detection.en.md
  92. 46 16
      docs/pipeline_usage/tutorials/cv_pipelines/object_detection.md
  93. 42 15
      docs/pipeline_usage/tutorials/cv_pipelines/open_vocabulary_detection.en.md
  94. 45 15
      docs/pipeline_usage/tutorials/cv_pipelines/open_vocabulary_detection.md
  95. 41 14
      docs/pipeline_usage/tutorials/cv_pipelines/open_vocabulary_segmentation.en.md
  96. 41 14
      docs/pipeline_usage/tutorials/cv_pipelines/open_vocabulary_segmentation.md
  97. 49 19
      docs/pipeline_usage/tutorials/cv_pipelines/pedestrian_attribute_recognition.en.md
  98. 50 19
      docs/pipeline_usage/tutorials/cv_pipelines/pedestrian_attribute_recognition.md
  99. 43 83
      docs/pipeline_usage/tutorials/cv_pipelines/rotated_object_detection.en.md
  100. 46 16
      docs/pipeline_usage/tutorials/cv_pipelines/rotated_object_detection.md

+ 58 - 15
docs/module_usage/tutorials/cv_modules/3d_bev_detection.en.md

@@ -26,21 +26,48 @@ The 3D multimodal fusion detection module is a key component in the fields of co
 <tr>
 </table>
 
-**Test Environment Description**:
-
-- **Performance Test Environment**
-  - **Test Dataset**: The above accuracy metrics are based on the <a href="https://www.nuscenes.org/nuscenes">nuscenes</a> validation set with mAP(0.5:0.95) and NDS 60.9, and the precision type is FP32.
-  - **Hardware Configuration**:
-    - GPU: NVIDIA Tesla T4
-    - CPU: Intel Xeon Gold 6271C @ 2.60GHz
-    - Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **Inference Mode Description**
-
-| Mode        | GPU Configuration                        | CPU Configuration | Acceleration Technology Combination                   |
-|-------------|----------------------------------------|-------------------|---------------------------------------------------|
-| Normal Mode | FP32 Precision / No TRT Acceleration   | FP32 Precision / 8 Threads | PaddleInference                                 |
-| High-Performance Mode | Optimal combination of pre-selected precision types and acceleration strategies | FP32 Precision / 8 Threads | Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.) |
+<strong>Test Environment Description:</strong>
+
+  <ul>
+      <li><b>Performance Test Environment</b>
+          <ul>
+          <li><strong>Test Dataset:</strong>The above accuracy metrics are based on the <a href="https://www.nuscenes.org/nuscenes">nuscenes</a> validation set with mAP(0.5:0.95) and NDS 60.9, and the precision type is FP32.</li>
+              <li><strong>Hardware Configuration:</strong>
+                  <ul>
+                      <li>GPU: NVIDIA Tesla T4</li>
+                      <li>CPU: Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>Inference Mode Description</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>Mode</th>
+            <th>GPU Configuration </th>
+            <th>CPU Configuration </th>
+            <th>Acceleration Technology Combination</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>Normal Mode</td>
+            <td>FP32 Precision / No TRT Acceleration</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>High-Performance Mode</td>
+            <td>Optimal combination of pre-selected precision types and acceleration strategies</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.)</td>
+        </tr>
+    </tbody>
+</table>
 
 ## III. Quick Integration
 > ❗ Before quick integration, please install the PaddleX wheel package first. For details, refer to the [PaddleX Local Installation Guide](../../../installation/installation.en.md).
@@ -70,6 +97,8 @@ pip install open3d
 python paddlex/inference/models/3d_bev_detection/visualizer_3d.py --save_path="./output/"
 ```
 
+<img src="https://raw.githubusercontent.com/cuicheng01/PaddleX_doc_images/main/images/pipelines/3d_bev_detection/02.png">
+
 After running, the result obtained is:
 
 ```bash
@@ -152,6 +181,20 @@ The following is an explanation of relevant methods and parameters:
 <td>No</td>
 <td>None</td>
 </tr>
+<tr>
+<td><code>device</code></td>
+<td>The device used for model inference</td>
+<td><code>str</code></td>
+<td>It supports specifying specific GPU card numbers, such as "gpu:0", other hardware card numbers, such as "npu:0", or CPU, such as "cpu".</td>
+<td><code>gpu:0</code></td>
+</tr>
+<tr>
+<td><code>use_hpip</code></td>
+<td>Whether to enable high-performance inference. </td>
+<td><code>bool</code></td>
+<td>None</td>
+<td><code>False</code></td>
+</tr>
 </table>
 
 * The `model_name` must be specified. After specifying `model_name`, the default model parameters built into PaddleX will be used. If `model_dir` is specified, the user-defined model will be used.

+ 58 - 17
docs/module_usage/tutorials/cv_modules/3d_bev_detection.md

@@ -30,21 +30,49 @@ comments: true
 
 </table>
 
-**测试环境说明:**
-
-- **性能测试环境**
-  - **测试数据集**:<a href="https://www.nuscenes.org/nuscenes">nuscenes</a>验证集 mAP(0.5:0.95), NDS 60.9, 精度类型为 FP32。
-  - **硬件配置**:
-    - GPU:NVIDIA Tesla T4
-    - CPU:Intel Xeon Gold 6271C @ 2.60GHz
-    - 其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **推理模式说明**
-
-| 模式        | GPU配置                          | CPU配置          | 加速技术组合                                |
-|-------------|----------------------------------|------------------|---------------------------------------------|
-| 常规模式    | FP32精度 / 无TRT加速             | FP32精度 / 8线程       | PaddleInference                             |
-| 高性能模式  | 选择先验精度类型和加速策略的最优组合         | FP32精度 / 8线程       | 选择先验最优后端(Paddle/OpenVINO/TRT等) |
+<strong>测试环境说明:</strong>
+
+  <ul>
+      <li><b>性能测试环境</b>
+          <ul>
+              <li><strong>测试数据集:</strong><a href="https://www.nuscenes.org/nuscenes">nuscenes</a>验证集 mAP(0.5:0.95), NDS 60.9, 精度类型为 FP32。</li>
+              <li><strong>硬件配置:</strong>
+                  <ul>
+                      <li>GPU:NVIDIA Tesla T4</li>
+                      <li>CPU:Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>推理模式说明</b></li>
+  </ul>
+
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>模式</th>
+            <th>GPU配置</th>
+            <th>CPU配置</th>
+            <th>加速技术组合</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>常规模式</td>
+            <td>FP32精度 / 无TRT加速</td>
+            <td>FP32精度 / 8线程</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>高性能模式</td>
+            <td>选择先验精度类型和加速策略的最优组合</td>
+            <td>FP32精度 / 8线程</td>
+            <td>选择先验最优后端(Paddle/OpenVINO/TRT等)</td>
+        </tr>
+    </tbody>
+</table>
 
 </details>
 
@@ -77,8 +105,7 @@ pip install open3d
 python paddlex/inference/models/3d_bev_detection/visualizer_3d.py --save_path="./output/"
 ```
 
-<img src="https://raw.githubusercontent.com/cuicheng01/PaddleX_doc_images/refs/heads/main/images/images/pipelines/3d_bev_detection/02.png">
-
+<img src="https://raw.githubusercontent.com/cuicheng01/PaddleX_doc_images/main/images/pipelines/3d_bev_detection/02.png">
 
 运行后,得到的结果为:
 ```bash
@@ -162,6 +189,20 @@ python paddlex/inference/models/3d_bev_detection/visualizer_3d.py --save_path=".
 <td>无</td>
 <td>无</td>
 </tr>
+<tr>
+<td><code>device</code></td>
+<td>模型推理设备</td>
+<td><code>str</code></td>
+<td>支持指定GPU具体卡号,如“gpu:0”,其他硬件具体卡号,如“npu:0”,CPU如“cpu”。</td>
+<td><code>gpu:0</code></td>
+</tr>
+<tr>
+<td><code>use_hpip</code></td>
+<td>是否启用高性能推理</td>
+<td><code>bool</code></td>
+<td>无</td>
+<td><code>False</code></td>
+</tr>
 </table>
 
 * 其中,`model_name` 必须指定,指定 `model_name` 后,默认使用 PaddleX 内置的模型参数,在此基础上,指定 `model_dir` 时,使用用户自定义的模型。

+ 55 - 15
docs/module_usage/tutorials/cv_modules/anomaly_detection.en.md

@@ -29,21 +29,47 @@ Unsupervised anomaly detection is a technology that automatically identifies and
 </tbody>
 </table>
 
-**Test Environment Description**:
-
-- **Performance Test Environment**
-  - **Test Dataset**: The above model accuracy indicators are measured from the MVTec_AD dataset.
-  - **Hardware Configuration**:
-    - GPU: NVIDIA Tesla T4
-    - CPU: Intel Xeon Gold 6271C @ 2.60GHz
-    - Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **Inference Mode Description**
-
-| Mode        | GPU Configuration                        | CPU Configuration | Acceleration Technology Combination                   |
-|-------------|----------------------------------------|-------------------|---------------------------------------------------|
-| Normal Mode | FP32 Precision / No TRT Acceleration   | FP32 Precision / 8 Threads | PaddleInference                                 |
-| High-Performance Mode | Optimal combination of pre-selected precision types and acceleration strategies | FP32 Precision / 8 Threads | Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.) |
+<strong>Test Environment Description:</strong>
+  <ul>
+      <li><b>Performance Test Environment</b>
+          <ul>
+              <li><strong>Test Dataset:</strong>The above model accuracy indicators are measured from the MVTec_AD dataset.</li>
+              <li><strong>Hardware Configuration:</strong>
+                  <ul>
+                      <li>GPU: NVIDIA Tesla T4</li>
+                      <li>CPU: Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>Inference Mode Description</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>Mode</th>
+            <th>GPU Configuration </th>
+            <th>CPU Configuration </th>
+            <th>Acceleration Technology Combination</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>Normal Mode</td>
+            <td>FP32 Precision / No TRT Acceleration</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>High-Performance Mode</td>
+            <td>Optimal combination of pre-selected precision types and acceleration strategies</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.)</td>
+        </tr>
+    </tbody>
+</table>
 
 ## III. Quick Integration  <a id="quick"> </a>
 Before quick integration, you need to install the PaddleX wheel package. For the installation method of the wheel package, please refer to the [PaddleX Local Installation Tutorial](../../../installation/installation.en.md). After installing the wheel package, a few lines of code can complete the inference of the unsupervised anomaly detection module. You can switch models under this module freely, and you can also integrate the model inference of the unsupervised anomaly detection module into your project. Before running the following code, please download the [demo image](https://paddle-model-ecology.bj.bcebos.com/paddlex/imgs/demo_image/uad_grid.png) to your local machine.
@@ -102,6 +128,20 @@ Relevant methods, parameters, and explanations are as follows:
 <td>None</td>
 <td>None</td>
 </tr>
+<tr>
+<td><code>device</code></td>
+<td>The device used for model inference</td>
+<td><code>str</code></td>
+<td>It supports specifying specific GPU card numbers, such as "gpu:0", other hardware card numbers, such as "npu:0", or CPU, such as "cpu".</td>
+<td><code>gpu:0</code></td>
+</tr>
+<tr>
+<td><code>use_hpip</code></td>
+<td>Whether to enable high-performance inference. </td>
+<td><code>bool</code></td>
+<td>None</td>
+<td><code>False</code></td>
+</tr>
 </table>
 
 * The `model_name` must be specified. After specifying `model_name`, the default model parameters built into PaddleX will be used. If `model_dir` is specified, the user-defined model will be used.

+ 49 - 15
docs/module_usage/tutorials/cv_modules/anomaly_detection.md

@@ -29,21 +29,48 @@ comments: true
 </tbody>
 </table>
 
-**测试环境说明:**
-
-- **性能测试环境**
-  - **测试数据集**:MVTec_AD 数据集中的grid类别。
-  - **硬件配置**:
-    - GPU:NVIDIA Tesla T4
-    - CPU:Intel Xeon Gold 6271C @ 2.60GHz
-    - 其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **推理模式说明**
-
-| 模式        | GPU配置                          | CPU配置          | 加速技术组合                                |
-|-------------|----------------------------------|------------------|---------------------------------------------|
-| 常规模式    | FP32精度 / 无TRT加速             | FP32精度 / 8线程       | PaddleInference                             |
-| 高性能模式  | 选择先验精度类型和加速策略的最优组合         | FP32精度 / 8线程       | 选择先验最优后端(Paddle/OpenVINO/TRT等) |
+<strong>测试环境说明:</strong>
+
+  <ul>
+      <li><b>性能测试环境</b>
+          <ul>
+              <li><strong>测试数据集:</strong>MVTec_AD 数据集中的grid类别。</li>
+              <li><strong>硬件配置:</strong>
+                  <ul>
+                      <li>GPU:NVIDIA Tesla T4</li>
+                      <li>CPU:Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>推理模式说明</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>模式</th>
+            <th>GPU配置</th>
+            <th>CPU配置</th>
+            <th>加速技术组合</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>常规模式</td>
+            <td>FP32精度 / 无TRT加速</td>
+            <td>FP32精度 / 8线程</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>高性能模式</td>
+            <td>选择先验精度类型和加速策略的最优组合</td>
+            <td>FP32精度 / 8线程</td>
+            <td>选择先验最优后端(Paddle/OpenVINO/TRT等)</td>
+        </tr>
+    </tbody>
+</table>
 
 
 ## 三、快速集成
@@ -107,6 +134,13 @@ for res in output:
 <td>无</td>
 </tr>
 <tr>
+<td><code>device</code></td>
+<td>模型推理设备</td>
+<td><code>str</code></td>
+<td>支持指定GPU具体卡号,如“gpu:0”,其他硬件具体卡号,如“npu:0”,CPU如“cpu”。</td>
+<td><code>gpu:0</code></td>
+</tr>
+<tr>
 <td><code>use_hpip</code></td>
 <td>是否启用高性能推理</td>
 <td><code>bool</code></td>

+ 56 - 15
docs/module_usage/tutorials/cv_modules/face_detection.en.md

@@ -57,21 +57,48 @@ Face detection is a fundamental task in object detection, aiming to automaticall
 </tbody>
 </table>
 
-**Test Environment Description**:
-
-- **Performance Test Environment**
-  - **Test Dataset**: The above accuracy metrics are evaluated on the WIDER-FACE validation set with an input size of 640*640.
-  - **Hardware Configuration**:
-    - GPU: NVIDIA Tesla T4
-    - CPU: Intel Xeon Gold 6271C @ 2.60GHz
-    - Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **Inference Mode Description**
-
-| Mode        | GPU Configuration                        | CPU Configuration | Acceleration Technology Combination                   |
-|-------------|----------------------------------------|-------------------|---------------------------------------------------|
-| Normal Mode | FP32 Precision / No TRT Acceleration   | FP32 Precision / 8 Threads | PaddleInference                                 |
-| High-Performance Mode | Optimal combination of pre-selected precision types and acceleration strategies | FP32 Precision / 8 Threads | Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.) |
+<strong>Test Environment Description:</strong>
+
+  <ul>
+      <li><b>Performance Test Environment</b>
+          <ul>
+                    <li><strong>Test Dataset:</strong>The above accuracy metrics are evaluated on the WIDER-FACE validation set with an input size of 640*640.</li>
+              <li><strong>Hardware Configuration:</strong>
+                  <ul>
+                      <li>GPU: NVIDIA Tesla T4</li>
+                      <li>CPU: Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>Inference Mode Description</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>Mode</th>
+            <th>GPU Configuration </th>
+            <th>CPU Configuration </th>
+            <th>Acceleration Technology Combination</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>Normal Mode</td>
+            <td>FP32 Precision / No TRT Acceleration</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>High-Performance Mode</td>
+            <td>Optimal combination of pre-selected precision types and acceleration strategies</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.)</td>
+        </tr>
+    </tbody>
+</table>
 
 
 ## III. Quick Integration  <a id="quick"> </a>
@@ -144,6 +171,13 @@ The explanations for the methods, parameters, etc., are as follows:
 <td>None</td>
 </tr>
 <tr>
+<td><code>device</code></td>
+<td>The device used for model inference</td>
+<td><code>str</code></td>
+<td>It supports specifying specific GPU card numbers, such as "gpu:0", other hardware card numbers, such as "npu:0", or CPU, such as "cpu".</td>
+<td><code>gpu:0</code></td>
+</tr>
+<tr>
 <td><code>img_size</code></td>
 <td>Size of the input image; if not specified, the default configuration of the PaddleX official model will be used</td>
 <td><code>int/list</code></td>
@@ -162,6 +196,13 @@ The explanations for the methods, parameters, etc., are as follows:
 <td>None</td>
 <td>None</td>
 </tr>
+<tr>
+<td><code>use_hpip</code></td>
+<td>Whether to enable high-performance inference. </td>
+<td><code>bool</code></td>
+<td>None</td>
+<td><code>False</code></td>
+</tr>
 </table>
 
 * The `model_name` must be specified. After specifying `model_name`, the default model parameters built into PaddleX are used. If `model_dir` is specified, the user-defined model is used.

+ 50 - 15
docs/module_usage/tutorials/cv_modules/face_detection.md

@@ -56,21 +56,49 @@ comments: true
 </tbody>
 </table>
 
-**测试环境说明:**
-
-- **性能测试环境**
-  - **测试数据集**:COCO 格式的 WIDER-FACE 验证集上,以640*640作为输入尺寸评估得到。
-  - **硬件配置**:
-    - GPU:NVIDIA Tesla T4
-    - CPU:Intel Xeon Gold 6271C @ 2.60GHz
-    - 其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **推理模式说明**
-
-| 模式        | GPU配置                          | CPU配置          | 加速技术组合                                |
-|-------------|----------------------------------|------------------|---------------------------------------------|
-| 常规模式    | FP32精度 / 无TRT加速             | FP32精度 / 8线程       | PaddleInference                             |
-| 高性能模式  | 选择先验精度类型和加速策略的最优组合         | FP32精度 / 8线程       | 选择先验最优后端(Paddle/OpenVINO/TRT等) |
+<strong>测试环境说明:</strong>
+
+  <ul>
+      <li><b>性能测试环境</b>
+          <ul>
+              <li><strong>测试数据集:</strong>COCO 格式的 WIDER-FACE 验证集上,以640*640作为输入尺寸评估得到。</li>
+              <li><strong>硬件配置:</strong>
+                  <ul>
+                      <li>GPU:NVIDIA Tesla T4</li>
+                      <li>CPU:Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>推理模式说明</b></li>
+  </ul>
+
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>模式</th>
+            <th>GPU配置</th>
+            <th>CPU配置</th>
+            <th>加速技术组合</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>常规模式</td>
+            <td>FP32精度 / 无TRT加速</td>
+            <td>FP32精度 / 8线程</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>高性能模式</td>
+            <td>选择先验精度类型和加速策略的最优组合</td>
+            <td>FP32精度 / 8线程</td>
+            <td>选择先验最优后端(Paddle/OpenVINO/TRT等)</td>
+        </tr>
+    </tbody>
+</table>
 
 
 ## 三、快速集成
@@ -138,6 +166,13 @@ for res in output:
 <td>无</td>
 </tr>
 <tr>
+<td><code>device</code></td>
+<td>模型推理设备</td>
+<td><code>str</code></td>
+<td>支持指定GPU具体卡号,如“gpu:0”,其他硬件具体卡号,如“npu:0”,CPU如“cpu”。</td>
+<td><code>gpu:0</code></td>
+</tr>
+<tr>
 <td><code>img_size</code></td>
 <td>输入图像大小;如果不指定,将默认使用PaddleX官方模型配置</td>
 <td><code>int/list/None</code></td>

+ 56 - 15
docs/module_usage/tutorials/cv_modules/face_feature.en.md

@@ -43,21 +43,48 @@ Face feature models typically take standardized face images processed through de
 </tbody>
 </table>
 
-**Test Environment Description**:
-
-- **Performance Test Environment**
-  - **Test Dataset**: The above accuracy metrics are Accuracy scores measured on the AgeDB-30, CFP-FP, and LFW datasets.
-  - **Hardware Configuration**:
-    - GPU: NVIDIA Tesla T4
-    - CPU: Intel Xeon Gold 6271C @ 2.60GHz
-    - Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **Inference Mode Description**
-
-| Mode        | GPU Configuration                        | CPU Configuration | Acceleration Technology Combination                   |
-|-------------|----------------------------------------|-------------------|---------------------------------------------------|
-| Normal Mode | FP32 Precision / No TRT Acceleration   | FP32 Precision / 8 Threads | PaddleInference                                 |
-| High-Performance Mode | Optimal combination of pre-selected precision types and acceleration strategies | FP32 Precision / 8 Threads | Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.) |
+<strong>Test Environment Description:</strong>
+
+  <ul>
+      <li><b>Performance Test Environment</b>
+          <ul>
+               <li><strong>Test Dataset:</strong>The above accuracy metrics are Accuracy scores measured on the AgeDB-30, CFP-FP, and LFW datasets.</li>
+              <li><strong>Hardware Configuration:</strong>
+                  <ul>
+                      <li>GPU: NVIDIA Tesla T4</li>
+                      <li>CPU: Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>Inference Mode Description</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>Mode</th>
+            <th>GPU Configuration </th>
+            <th>CPU Configuration </th>
+            <th>Acceleration Technology Combination</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>Normal Mode</td>
+            <td>FP32 Precision / No TRT Acceleration</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>High-Performance Mode</td>
+            <td>Optimal combination of pre-selected precision types and acceleration strategies</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.)</td>
+        </tr>
+    </tbody>
+</table>
 
 ## III. Quick Integration
 > ❗ Before quick integration, please install the PaddleX wheel package. For details, refer to the [PaddleX Local Installation Tutorial](../../../installation/installation.en.md)
@@ -116,12 +143,26 @@ The explanations for the methods, parameters, etc., are as follows:
 <td>None</td>
 </tr>
 <tr>
+<td><code>device</code></td>
+<td>The device used for model inference</td>
+<td><code>str</code></td>
+<td>It supports specifying specific GPU card numbers, such as "gpu:0", other hardware card numbers, such as "npu:0", or CPU, such as "cpu".</td>
+<td><code>gpu:0</code></td>
+</tr>
+<tr>
 <td><code>flip</code></td>
 <td>Whether to perform flipped inference; if True, the model will infer the horizontally flipped input image and fuse the results of both inferences to improve the accuracy of face features</td>
 <td><code>bool</code></td>
 <td>None</td>
 <td><code>False</code></td>
 </tr>
+<tr>
+<td><code>use_hpip</code></td>
+<td>Whether to enable high-performance inference. </td>
+<td><code>bool</code></td>
+<td>None</td>
+<td><code>False</code></td>
+</tr>
 </table>
 
 * The `model_name` must be specified. After specifying `model_name`, the default model parameters built into PaddleX are used. If `model_dir` is specified, the user-defined model is used.

+ 49 - 15
docs/module_usage/tutorials/cv_modules/face_feature.md

@@ -44,21 +44,48 @@ comments: true
 </tbody>
 </table>
 
-**测试环境说明:**
-
-- **性能测试环境**
-  - **测试数据集**:分别在AgeDB-30、CFP-FP和LFW数据集上测得。
-  - **硬件配置**:
-    - GPU:NVIDIA Tesla T4
-    - CPU:Intel Xeon Gold 6271C @ 2.60GHz
-    - 其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **推理模式说明**
-
-| 模式        | GPU配置                          | CPU配置          | 加速技术组合                                |
-|-------------|----------------------------------|------------------|---------------------------------------------|
-| 常规模式    | FP32精度 / 无TRT加速             | FP32精度 / 8线程       | PaddleInference                             |
-| 高性能模式  | 选择先验精度类型和加速策略的最优组合         | FP32精度 / 8线程       | 选择先验最优后端(Paddle/OpenVINO/TRT等) |
+<strong>测试环境说明:</strong>
+
+  <ul>
+      <li><b>性能测试环境</b>
+          <ul>
+              <li><strong>测试数据集:</strong>分别在AgeDB-30、CFP-FP和LFW数据集上测得。</li>
+              <li><strong>硬件配置:</strong>
+                  <ul>
+                      <li>GPU:NVIDIA Tesla T4</li>
+                      <li>CPU:Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>推理模式说明</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>模式</th>
+            <th>GPU配置</th>
+            <th>CPU配置</th>
+            <th>加速技术组合</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>常规模式</td>
+            <td>FP32精度 / 无TRT加速</td>
+            <td>FP32精度 / 8线程</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>高性能模式</td>
+            <td>选择先验精度类型和加速策略的最优组合</td>
+            <td>FP32精度 / 8线程</td>
+            <td>选择先验最优后端(Paddle/OpenVINO/TRT等)</td>
+        </tr>
+    </tbody>
+</table>
 
 ## 三、快速集成
 > ❗ 在快速集成前,请先安装 PaddleX 的 wheel 包,详细请参考 [PaddleX本地安装教程](../../../installation/installation.md)
@@ -141,6 +168,13 @@ for res in output:
 <td>无</td>
 </tr>
 <tr>
+<td><code>device</code></td>
+<td>模型推理设备</td>
+<td><code>str</code></td>
+<td>支持指定GPU具体卡号,如“gpu:0”,其他硬件具体卡号,如“npu:0”,CPU如“cpu”。</td>
+<td><code>gpu:0</code></td>
+</tr>
+<tr>
 <td><code>flip</code></td>
 <td>是否进行反转推理; 如果为True,模型会对输入图像水平翻转后再次推理,并融合两次推理结果以提升人脸特征的准确性</td>
 <td><code>bool</code></td>

+ 56 - 15
docs/module_usage/tutorials/cv_modules/human_detection.en.md

@@ -39,21 +39,48 @@ Human detection is a subtask of object detection, which utilizes computer vision
 </tr>
 </table>
 
-**Test Environment Description**:
-
-- **Performance Test Environment**
-  - **Test Dataset**: The evaluation set for the above accuracy metrics is CrowdHuman dataset mAP(0.5:0.95).
-  - **Hardware Configuration**:
-    - GPU: NVIDIA Tesla T4
-    - CPU: Intel Xeon Gold 6271C @ 2.60GHz
-    - Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **Inference Mode Description**
-
-| Mode        | GPU Configuration                        | CPU Configuration | Acceleration Technology Combination                   |
-|-------------|----------------------------------------|-------------------|---------------------------------------------------|
-| Normal Mode | FP32 Precision / No TRT Acceleration   | FP32 Precision / 8 Threads | PaddleInference                                 |
-| High-Performance Mode | Optimal combination of pre-selected precision types and acceleration strategies | FP32 Precision / 8 Threads | Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.) |
+<strong>Test Environment Description:</strong>
+
+  <ul>
+      <li><b>Performance Test Environment</b>
+          <ul>
+                <li><strong>Test Dataset:</strong>The evaluation set for the above accuracy metrics is CrowdHuman dataset mAP(0.5:0.95).</li>
+              <li><strong>Hardware Configuration:</strong>
+                  <ul>
+                      <li>GPU: NVIDIA Tesla T4</li>
+                      <li>CPU: Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>Inference Mode Description</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>Mode</th>
+            <th>GPU Configuration </th>
+            <th>CPU Configuration </th>
+            <th>Acceleration Technology Combination</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>Normal Mode</td>
+            <td>FP32 Precision / No TRT Acceleration</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>High-Performance Mode</td>
+            <td>Optimal combination of pre-selected precision types and acceleration strategies</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.)</td>
+        </tr>
+    </tbody>
+</table>
 
 
 ## III. Quick Integration
@@ -120,12 +147,26 @@ The explanations for the methods, parameters, etc., are as follows:
 <td>None</td>
 </tr>
 <tr>
+<td><code>device</code></td>
+<td>The device used for model inference</td>
+<td><code>str</code></td>
+<td>It supports specifying specific GPU card numbers, such as "gpu:0", other hardware card numbers, such as "npu:0", or CPU, such as "cpu".</td>
+<td><code>gpu:0</code></td>
+</tr>
+<tr>
 <td><code>threshold</code></td>
 <td>Threshold for filtering low-confidence objects</td>
 <td><code>float/None/dict</code></td>
 <td>None</td>
 <td>None</td>
 </tr>
+<tr>
+<td><code>use_hpip</code></td>
+<td>Whether to enable high-performance inference. </td>
+<td><code>bool</code></td>
+<td>None</td>
+<td><code>False</code></td>
+</tr>
 </table>
 
 * The `model_name` must be specified. After specifying `model_name`, the default model parameters built into PaddleX are used. If `model_dir` is specified, the user-defined model is used.

+ 49 - 15
docs/module_usage/tutorials/cv_modules/human_detection.md

@@ -38,21 +38,48 @@ comments: true
 </tr>
 </table>
 
-**测试环境说明:**
-
-- **性能测试环境**
-  - **测试数据集**:CrowdHuman数据集 mAP(0.5:0.95)。
-  - **硬件配置**:
-    - GPU:NVIDIA Tesla T4
-    - CPU:Intel Xeon Gold 6271C @ 2.60GHz
-    - 其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **推理模式说明**
-
-| 模式        | GPU配置                          | CPU配置          | 加速技术组合                                |
-|-------------|----------------------------------|------------------|---------------------------------------------|
-| 常规模式    | FP32精度 / 无TRT加速             | FP32精度 / 8线程       | PaddleInference                             |
-| 高性能模式  | 选择先验精度类型和加速策略的最优组合         | FP32精度 / 8线程       | 选择先验最优后端(Paddle/OpenVINO/TRT等) |
+<strong>测试环境说明:</strong>
+
+  <ul>
+      <li><b>性能测试环境</b>
+          <ul>
+          <li><strong>测试数据集:</strong>CrowdHuman数据集 mAP(0.5:0.95)。</li>
+              <li><strong>硬件配置:</strong>
+                  <ul>
+                      <li>GPU:NVIDIA Tesla T4</li>
+                      <li>CPU:Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>推理模式说明</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>模式</th>
+            <th>GPU配置</th>
+            <th>CPU配置</th>
+            <th>加速技术组合</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>常规模式</td>
+            <td>FP32精度 / 无TRT加速</td>
+            <td>FP32精度 / 8线程</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>高性能模式</td>
+            <td>选择先验精度类型和加速策略的最优组合</td>
+            <td>FP32精度 / 8线程</td>
+            <td>选择先验最优后端(Paddle/OpenVINO/TRT等)</td>
+        </tr>
+    </tbody>
+</table>
 
 
 ## 三、快速集成
@@ -115,6 +142,13 @@ for res in output:
 <td>无</td>
 </tr>
 <tr>
+<td><code>device</code></td>
+<td>模型推理设备</td>
+<td><code>str</code></td>
+<td>支持指定GPU具体卡号,如“gpu:0”,其他硬件具体卡号,如“npu:0”,CPU如“cpu”。</td>
+<td><code>gpu:0</code></td>
+</tr>
+<tr>
 <td><code>threshold</code></td>
 <td>低分object过滤阈值</td>
 <td><code>float/None/dict[int, float]</code></td>

+ 56 - 15
docs/module_usage/tutorials/cv_modules/human_keypoint_detection.en.md

@@ -43,21 +43,48 @@ Keypoint detection algorithms mainly include two approaches: Top-Down and Bottom
   </tr>
 </table>
 
-**Test Environment Description**:
-
-- **Performance Test Environment**
-  - **Test Dataset**: The above accuracy metrics are based on the COCO dataset AP(0.5:0.95) using ground truth annotations for bounding boxes.
-  - **Hardware Configuration**:
-    - GPU: NVIDIA Tesla T4
-    - CPU: Intel Xeon Gold 6271C @ 2.60GHz
-    - Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **Inference Mode Description**
-
-| Mode        | GPU Configuration                        | CPU Configuration | Acceleration Technology Combination                   |
-|-------------|----------------------------------------|-------------------|---------------------------------------------------|
-| Normal Mode | FP32 Precision / No TRT Acceleration   | FP32 Precision / 8 Threads | PaddleInference                                 |
-| High-Performance Mode | Optimal combination of pre-selected precision types and acceleration strategies | FP32 Precision / 8 Threads | Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.) |
+<strong>Test Environment Description:</strong>
+
+  <ul>
+      <li><b>Performance Test Environment</b>
+          <ul>
+           <li><strong>Test Dataset:</strong>The above accuracy metrics are based on the COCO dataset AP(0.5:0.95) using ground truth annotations for bounding boxes.</li>
+              <li><strong>Hardware Configuration:</strong>
+                  <ul>
+                      <li>GPU: NVIDIA Tesla T4</li>
+                      <li>CPU: Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>Inference Mode Description</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>Mode</th>
+            <th>GPU Configuration </th>
+            <th>CPU Configuration </th>
+            <th>Acceleration Technology Combination</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>Normal Mode</td>
+            <td>FP32 Precision / No TRT Acceleration</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>High-Performance Mode</td>
+            <td>Optimal combination of pre-selected precision types and acceleration strategies</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.)</td>
+        </tr>
+    </tbody>
+</table>
 
 ## III. Quick Integration
 > ❗ Before quick integration, please install the PaddleX wheel package first. For details, please refer to the [PaddleX Local Installation Guide](../../../installation/installation.en.md)
@@ -124,12 +151,26 @@ The explanations for the methods, parameters, etc., are as follows:
 <td>None</td>
 </tr>
 <tr>
+<td><code>device</code></td>
+<td>The device used for model inference</td>
+<td><code>str</code></td>
+<td>It supports specifying specific GPU card numbers, such as "gpu:0", other hardware card numbers, such as "npu:0", or CPU, such as "cpu".</td>
+<td><code>gpu:0</code></td>
+</tr>
+<tr>
 <td><code>flip</code></td>
 <td>Whether to perform flipped inference; if True, the model will infer the horizontally flipped input image and fuse the results of both inferences to increase the accuracy of keypoint predictions</td>
 <td><code>bool</code></td>
 <td>None</td>
 <td><code>False</code></td>
 </tr>
+<tr>
+<td><code>use_hpip</code></td>
+<td>Whether to enable high-performance inference. </td>
+<td><code>bool</code></td>
+<td>None</td>
+<td><code>False</code></td>
+</tr>
 </table>
 
 * The `model_name` must be specified. After specifying `model_name`, the default model parameters built into PaddleX are used. If `model_dir` is specified, the user-defined model is used.

+ 56 - 15
docs/module_usage/tutorials/cv_modules/human_keypoint_detection.md

@@ -43,21 +43,48 @@ comments: true
   </tr>
 </table>
 
-**测试环境说明:**
-
-- **性能测试环境**
-  - **测试数据集**:COCO数据集 AP(0.5:0.95),所依赖的检测框为ground truth标注得到。
-  - **硬件配置**:
-    - GPU:NVIDIA Tesla T4
-    - CPU:Intel Xeon Gold 6271C @ 2.60GHz
-    - 其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **推理模式说明**
-
-| 模式        | GPU配置                          | CPU配置          | 加速技术组合                                |
-|-------------|----------------------------------|------------------|---------------------------------------------|
-| 常规模式    | FP32精度 / 无TRT加速             | FP32精度 / 8线程       | PaddleInference                             |
-| 高性能模式  | 选择先验精度类型和加速策略的最优组合         | FP32精度 / 8线程       | 选择先验最优后端(Paddle/OpenVINO/TRT等) |
+<strong>测试环境说明:</strong>
+
+  <ul>
+      <li><b>性能测试环境</b>
+          <ul>
+             <li><strong>测试数据集:</strong>COCO数据集 AP(0.5:0.95),所依赖的检测框为ground truth标注得到。</li>
+              <li><strong>硬件配置:</strong>
+                  <ul>
+                      <li>GPU:NVIDIA Tesla T4</li>
+                      <li>CPU:Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>推理模式说明</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>模式</th>
+            <th>GPU配置</th>
+            <th>CPU配置</th>
+            <th>加速技术组合</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>常规模式</td>
+            <td>FP32精度 / 无TRT加速</td>
+            <td>FP32精度 / 8线程</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>高性能模式</td>
+            <td>选择先验精度类型和加速策略的最优组合</td>
+            <td>FP32精度 / 8线程</td>
+            <td>选择先验最优后端(Paddle/OpenVINO/TRT等)</td>
+        </tr>
+    </tbody>
+</table>
 
 
 ## 三、快速集成
@@ -139,6 +166,20 @@ for res in output:
 <td>无</td>
 </tr>
 <tr>
+<td><code>device</code></td>
+<td>模型推理设备</td>
+<td><code>str</code></td>
+<td>支持指定GPU具体卡号,如“gpu:0”,其他硬件具体卡号,如“npu:0”,CPU如“cpu”。</td>
+<td><code>gpu:0</code></td>
+</tr>
+<tr>
+<td><code>use_hpip</code></td>
+<td>是否启用高性能推理</td>
+<td><code>bool</code></td>
+<td>无</td>
+<td><code>False</code></td>
+</tr>
+<tr>
 <td><code>flip</code></td>
 <td>是否进行图像水平反转推理结果融合; 如果为True,模型会对输入图像水平翻转后再次推理,并融合两次推理结果以增加关键点预测的准确性</td>
 <td><code>bool</code></td>

+ 54 - 13
docs/module_usage/tutorials/cv_modules/image_classification.en.md

@@ -678,21 +678,48 @@ The image classification module is a crucial component in computer vision system
 </tr>
 </tr></tr></tr></tr></table>
 
-**Test Environment Description**:
+<strong>Test Environment Description:</strong>
 
-- **Performance Test Environment**
-  - **Test Dataset**: <a href="https://www.image-net.org/index.php">ImageNet-1k</a> validation set.
-  - **Hardware Configuration**:
-    - GPU: NVIDIA Tesla T4
-    - CPU: Intel Xeon Gold 6271C @ 2.60GHz
-    - Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
+  <ul>
+      <li><b>Performance Test Environment</b>
+          <ul>
+                <li><strong>Test Dataset:</strong><a href="https://www.image-net.org/index.php">ImageNet-1k</a> validation set.</li>
+              <li><strong>Hardware Configuration:</strong>
+                  <ul>
+                      <li>GPU: NVIDIA Tesla T4</li>
+                      <li>CPU: Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>Inference Mode Description</b></li>
+  </ul>
 
-- **Inference Mode Description**
-
-| Mode        | GPU Configuration                        | CPU Configuration | Acceleration Technology Combination                   |
-|-------------|----------------------------------------|-------------------|---------------------------------------------------|
-| Normal Mode | FP32 Precision / No TRT Acceleration   | FP32 Precision / 8 Threads | PaddleInference                                 |
-| High-Performance Mode | Optimal combination of pre-selected precision types and acceleration strategies | FP32 Precision / 8 Threads | Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.) |
+<table border="1">
+    <thead>
+        <tr>
+            <th>Mode</th>
+            <th>GPU Configuration </th>
+            <th>CPU Configuration </th>
+            <th>Acceleration Technology Combination</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>Normal Mode</td>
+            <td>FP32 Precision / No TRT Acceleration</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>High-Performance Mode</td>
+            <td>Optimal combination of pre-selected precision types and acceleration strategies</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.)</td>
+        </tr>
+    </tbody>
+</table>
 
 </details>
 
@@ -757,6 +784,20 @@ Related methods, parameters, and explanations are as follows:
 <td>None</td>
 <td>None</td>
 </tr>
+<tr>
+<td><code>device</code></td>
+<td>The device used for model inference</td>
+<td><code>str</code></td>
+<td>It supports specifying specific GPU card numbers, such as "gpu:0", other hardware card numbers, such as "npu:0", or CPU, such as "cpu".</td>
+<td><code>gpu:0</code></td>
+</tr>
+<tr>
+<td><code>use_hpip</code></td>
+<td>Whether to enable high-performance inference. </td>
+<td><code>bool</code></td>
+<td>None</td>
+<td><code>False</code></td>
+</tr>
 </table>
 
 * The `model_name` must be specified. After specifying `model_name`, the default model parameters built into PaddleX are used. If `model_dir` is specified, the user-defined model is used.

+ 47 - 13
docs/module_usage/tutorials/cv_modules/image_classification.md

@@ -673,21 +673,48 @@ comments: true
 </tr>
 </table>
 
-**测试环境说明:**
+<strong>测试环境说明:</strong>
 
-- **性能测试环境**
-  - **测试数据集**:<a href="https://www.image-net.org/index.php">ImageNet-1k</a> 验证集 Top1 Acc。
-  - **硬件配置**:
-    - GPU:NVIDIA Tesla T4
-    - CPU:Intel Xeon Gold 6271C @ 2.60GHz
-    - 其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
+  <ul>
+      <li><b>性能测试环境</b>
+          <ul>
+           <li><strong>测试数据集:</strong><a href="https://www.image-net.org/index.php">ImageNet-1k</a> 验证集 Top1 Acc。</li>
+              <li><strong>硬件配置:</strong>
+                  <ul>
+                      <li>GPU:NVIDIA Tesla T4</li>
+                      <li>CPU:Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>推理模式说明</b></li>
+  </ul>
 
-- **推理模式说明**
-
-| 模式        | GPU配置                          | CPU配置          | 加速技术组合                                |
-|-------------|----------------------------------|------------------|---------------------------------------------|
-| 常规模式    | FP32精度 / 无TRT加速             | FP32精度 / 8线程       | PaddleInference                             |
-| 高性能模式  | 选择先验精度类型和加速策略的最优组合         | FP32精度 / 8线程       | 选择先验最优后端(Paddle/OpenVINO/TRT等) |
+<table border="1">
+    <thead>
+        <tr>
+            <th>模式</th>
+            <th>GPU配置</th>
+            <th>CPU配置</th>
+            <th>加速技术组合</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>常规模式</td>
+            <td>FP32精度 / 无TRT加速</td>
+            <td>FP32精度 / 8线程</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>高性能模式</td>
+            <td>选择先验精度类型和加速策略的最优组合</td>
+            <td>FP32精度 / 8线程</td>
+            <td>选择先验最优后端(Paddle/OpenVINO/TRT等)</td>
+        </tr>
+    </tbody>
+</table>
 
 </details>
 
@@ -750,6 +777,13 @@ for res in output:
 <td>无</td>
 </tr>
 <tr>
+<td><code>device</code></td>
+<td>模型推理设备</td>
+<td><code>str</code></td>
+<td>支持指定GPU具体卡号,如“gpu:0”,其他硬件具体卡号,如“npu:0”,CPU如“cpu”。</td>
+<td><code>gpu:0</code></td>
+</tr>
+<tr>
 <td><code>topk</code></td>
 <td>预测结果的前<code>topk</code>值,如果不指定,将默认使用PaddleX官方模型配置</td>
 <td><code>int</code></td>

+ 56 - 15
docs/module_usage/tutorials/cv_modules/image_feature.en.md

@@ -43,21 +43,48 @@ The image feature module is one of the important tasks in computer vision, prima
 </tr>
 </table>
 
-**Test Environment Description**:
-
-- **Performance Test Environment**
-  - **Test Dataset**: PaddleX Custom Dataset Creation.
-  - **Hardware Configuration**:
-    - GPU: NVIDIA Tesla T4
-    - CPU: Intel Xeon Gold 6271C @ 2.60GHz
-    - Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **Inference Mode Description**
-
-| Mode        | GPU Configuration                        | CPU Configuration | Acceleration Technology Combination                   |
-|-------------|----------------------------------------|-------------------|---------------------------------------------------|
-| Normal Mode | FP32 Precision / No TRT Acceleration   | FP32 Precision / 8 Threads | PaddleInference                                 |
-| High-Performance Mode | Optimal combination of pre-selected precision types and acceleration strategies | FP32 Precision / 8 Threads | Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.) |
+<strong>Test Environment Description:</strong>
+
+  <ul>
+      <li><b>Performance Test Environment</b>
+          <ul>
+              <li><strong>Test Dataset:</strong>PaddleX Custom Dataset.</li>
+              <li><strong>Hardware Configuration:</strong>
+                  <ul>
+                      <li>GPU: NVIDIA Tesla T4</li>
+                      <li>CPU: Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>Inference Mode Description</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>Mode</th>
+            <th>GPU Configuration </th>
+            <th>CPU Configuration </th>
+            <th>Acceleration Technology Combination</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>Normal Mode</td>
+            <td>FP32 Precision / No TRT Acceleration</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>High-Performance Mode</td>
+            <td>Optimal combination of pre-selected precision types and acceleration strategies</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.)</td>
+        </tr>
+    </tbody>
+</table>
 
 ## III. Quick Integration
 > ❗ Before quick integration, please install the PaddleX wheel package. For detailed instructions, refer to the [PaddleX Local Installation Guide](../../../installation/installation.en.md)
@@ -112,6 +139,20 @@ Descriptions of related methods, parameters, etc., are as follows:
 <td>None</td>
 <td>None</td>
 </tr>
+<tr>
+<td><code>device</code></td>
+<td>The device used for model inference</td>
+<td><code>str</code></td>
+<td>It supports specifying specific GPU card numbers, such as "gpu:0", other hardware card numbers, such as "npu:0", or CPU, such as "cpu".</td>
+<td><code>gpu:0</code></td>
+</tr>
+<tr>
+<td><code>use_hpip</code></td>
+<td>Whether to enable high-performance inference. </td>
+<td><code>bool</code></td>
+<td>None</td>
+<td><code>False</code></td>
+</tr>
 </table>
 
 * The `model_name` must be specified. When `model_name` is specified, PaddleX's built-in model parameters are used by default. If `model_dir` is specified, the user-defined model is used.

+ 49 - 15
docs/module_usage/tutorials/cv_modules/image_feature.md

@@ -43,21 +43,48 @@ comments: true
 </tr>
 </table>
 
-**测试环境说明:**
-
-- **性能测试环境**
-  - **测试数据集**:PaddleX自建数据集。
-  - **硬件配置**:
-    - GPU:NVIDIA Tesla T4
-    - CPU:Intel Xeon Gold 6271C @ 2.60GHz
-    - 其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **推理模式说明**
-
-| 模式        | GPU配置                          | CPU配置          | 加速技术组合                                |
-|-------------|----------------------------------|------------------|---------------------------------------------|
-| 常规模式    | FP32精度 / 无TRT加速             | FP32精度 / 8线程       | PaddleInference                             |
-| 高性能模式  | 选择先验精度类型和加速策略的最优组合         | FP32精度 / 8线程       | 选择先验最优后端(Paddle/OpenVINO/TRT等) |
+<strong>测试环境说明:</strong>
+
+  <ul>
+      <li><b>性能测试环境</b>
+          <ul>
+            <li><strong>测试数据集:</strong>PaddleX自建数据集。</li>
+              <li><strong>硬件配置:</strong>
+                  <ul>
+                      <li>GPU:NVIDIA Tesla T4</li>
+                      <li>CPU:Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>推理模式说明</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>模式</th>
+            <th>GPU配置</th>
+            <th>CPU配置</th>
+            <th>加速技术组合</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>常规模式</td>
+            <td>FP32精度 / 无TRT加速</td>
+            <td>FP32精度 / 8线程</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>高性能模式</td>
+            <td>选择先验精度类型和加速策略的最优组合</td>
+            <td>FP32精度 / 8线程</td>
+            <td>选择先验最优后端(Paddle/OpenVINO/TRT等)</td>
+        </tr>
+    </tbody>
+</table>
 
 ## 三、快速集成
 > ❗ 在快速集成前,请先安装 PaddleX 的 wheel 包,详细请参考 [PaddleX本地安装教程](../../../installation/installation.md)
@@ -112,6 +139,13 @@ for res in output:
 <td>无</td>
 </tr>
 <tr>
+<td><code>device</code></td>
+<td>模型推理设备</td>
+<td><code>str</code></td>
+<td>支持指定GPU具体卡号,如“gpu:0”,其他硬件具体卡号,如“npu:0”,CPU如“cpu”。</td>
+<td><code>gpu:0</code></td>
+</tr>
+<tr>
 <td><code>use_hpip</code></td>
 <td>是否启用高性能推理</td>
 <td><code>bool</code></td>

+ 56 - 15
docs/module_usage/tutorials/cv_modules/image_multilabel_classification.en.md

@@ -67,21 +67,48 @@ The image multi-label classification module is a crucial component in computer v
 </tr>
 </table>
 
-**Test Environment Description**:
-
-- **Performance Test Environment**
-  - **Test Dataset**:  multi-label classification task on [COCO2017](https://cocodataset.org/#home).
-  - **Hardware Configuration**:
-    - GPU: NVIDIA Tesla T4
-    - CPU: Intel Xeon Gold 6271C @ 2.60GHz
-    - Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **Inference Mode Description**
-
-| Mode        | GPU Configuration                        | CPU Configuration | Acceleration Technology Combination                   |
-|-------------|----------------------------------------|-------------------|---------------------------------------------------|
-| Normal Mode | FP32 Precision / No TRT Acceleration   | FP32 Precision / 8 Threads | PaddleInference                                 |
-| High-Performance Mode | Optimal combination of pre-selected precision types and acceleration strategies | FP32 Precision / 8 Threads | Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.) |
+<strong>Test Environment Description:</strong>
+
+  <ul>
+      <li><b>Performance Test Environment</b>
+          <ul>
+          <li><strong>Test Dataset:</strong>multi-label classification task on  <a href="https://cocodataset.org/#home">COCO2017</a></li>
+              <li><strong>Hardware Configuration:</strong>
+                  <ul>
+                      <li>GPU: NVIDIA Tesla T4</li>
+                      <li>CPU: Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>Inference Mode Description</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>Mode</th>
+            <th>GPU Configuration </th>
+            <th>CPU Configuration </th>
+            <th>Acceleration Technology Combination</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>Normal Mode</td>
+            <td>FP32 Precision / No TRT Acceleration</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>High-Performance Mode</td>
+            <td>Optimal combination of pre-selected precision types and acceleration strategies</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.)</td>
+        </tr>
+    </tbody>
+</table>
 
 ## III. Quick Integration
 > ❗ Before quick integration, please install the PaddleX wheel package. For detailed instructions, refer to the [PaddleX Local Installation Guide](../../../installation/installation.en.md)
@@ -144,6 +171,20 @@ Related methods, parameters, and explanations are as follows:
 <td>None</td>
 <td>None</td>
 </tr>
+<tr>
+<td><code>device</code></td>
+<td>The device used for model inference</td>
+<td><code>str</code></td>
+<td>It supports specifying specific GPU card numbers, such as "gpu:0", other hardware card numbers, such as "npu:0", or CPU, such as "cpu".</td>
+<td><code>gpu:0</code></td>
+</tr>
+<tr>
+<td><code>use_hpip</code></td>
+<td>Whether to enable high-performance inference. </td>
+<td><code>bool</code></td>
+<td>None</td>
+<td><code>False</code></td>
+</tr>
 </table>
 
 * The `model_name` must be specified. After specifying `model_name`, the default model parameters built into PaddleX are used. If `model_dir` is specified, the user-defined model is used.

+ 49 - 15
docs/module_usage/tutorials/cv_modules/image_multilabel_classification.md

@@ -67,21 +67,48 @@ comments: true
 </tr>
 </table>
 
-**测试环境说明:**
-
-- **性能测试环境**
-  - **测试数据集**:[COCO2017](https://cocodataset.org/#home)的多标签分类任务。
-  - **硬件配置**:
-    - GPU:NVIDIA Tesla T4
-    - CPU:Intel Xeon Gold 6271C @ 2.60GHz
-    - 其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **推理模式说明**
-
-| 模式        | GPU配置                          | CPU配置          | 加速技术组合                                |
-|-------------|----------------------------------|------------------|---------------------------------------------|
-| 常规模式    | FP32精度 / 无TRT加速             | FP32精度 / 8线程       | PaddleInference                             |
-| 高性能模式  | 选择先验精度类型和加速策略的最优组合         | FP32精度 / 8线程       | 选择先验最优后端(Paddle/OpenVINO/TRT等) |
+<strong>测试环境说明:</strong>
+
+  <ul>
+      <li><b>性能测试环境</b>
+          <ul>
+          <li><strong>测试数据集:</strong><a href="https://cocodataset.org/#home">COCO2017</a> </li>
+              <li><strong>硬件配置:</strong>
+                  <ul>
+                      <li>GPU:NVIDIA Tesla T4</li>
+                      <li>CPU:Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>推理模式说明</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>模式</th>
+            <th>GPU配置</th>
+            <th>CPU配置</th>
+            <th>加速技术组合</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>常规模式</td>
+            <td>FP32精度 / 无TRT加速</td>
+            <td>FP32精度 / 8线程</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>高性能模式</td>
+            <td>选择先验精度类型和加速策略的最优组合</td>
+            <td>FP32精度 / 8线程</td>
+            <td>选择先验最优后端(Paddle/OpenVINO/TRT等)</td>
+        </tr>
+    </tbody>
+</table>
 
 ## 三、快速集成
  > ❗ 在快速集成前,请先安装 PaddleX 的 wheel 包,详细请参考 [PaddleX本地安装教程](../../../installation/installation.md)
@@ -142,6 +169,13 @@ for res in output:
 <td>无</td>
 </tr>
 <tr>
+<td><code>device</code></td>
+<td>模型推理设备</td>
+<td><code>str</code></td>
+<td>支持指定GPU具体卡号,如“gpu:0”,其他硬件具体卡号,如“npu:0”,CPU如“cpu”。</td>
+<td><code>gpu:0</code></td>
+</tr>
+<tr>
 <td><code>threshold</code></td>
 <td>多标签分类阈值</td>
 <td><code>float/list/dict</code></td>

+ 56 - 15
docs/module_usage/tutorials/cv_modules/instance_segmentation.en.md

@@ -160,21 +160,48 @@ The instance segmentation module is a crucial component in computer vision syste
 </tr>
 </table>
 
-**Test Environment Description**:
-
-- **Performance Test Environment**
-  - **Test Dataset**: <a href="https://cocodataset.org/#home">COCO2017</a> validation set.
-  - **Hardware Configuration**:
-    - GPU: NVIDIA Tesla T4
-    - CPU: Intel Xeon Gold 6271C @ 2.60GHz
-    - Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **Inference Mode Description**
-
-| Mode        | GPU Configuration                        | CPU Configuration | Acceleration Technology Combination                   |
-|-------------|----------------------------------------|-------------------|---------------------------------------------------|
-| Normal Mode | FP32 Precision / No TRT Acceleration   | FP32 Precision / 8 Threads | PaddleInference                                 |
-| High-Performance Mode | Optimal combination of pre-selected precision types and acceleration strategies | FP32 Precision / 8 Threads | Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.) |
+<strong>Test Environment Description:</strong>
+
+  <ul>
+      <li><b>Performance Test Environment</b>
+          <ul>
+           <li><strong>Test Dataset:</strong><a href="https://cocodataset.org/#home">COCO2017</a> validation set.</li>
+              <li><strong>Hardware Configuration:</strong>
+                  <ul>
+                      <li>GPU: NVIDIA Tesla T4</li>
+                      <li>CPU: Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>Inference Mode Description</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>Mode</th>
+            <th>GPU Configuration </th>
+            <th>CPU Configuration </th>
+            <th>Acceleration Technology Combination</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>Normal Mode</td>
+            <td>FP32 Precision / No TRT Acceleration</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>High-Performance Mode</td>
+            <td>Optimal combination of pre-selected precision types and acceleration strategies</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.)</td>
+        </tr>
+    </tbody>
+</table>
 
 </details>
 
@@ -243,12 +270,26 @@ Related methods, parameters, and explanations are as follows:
 <td>None</td>
 </tr>
 <tr>
+<td><code>device</code></td>
+<td>The device used for model inference</td>
+<td><code>str</code></td>
+<td>It supports specifying specific GPU card numbers, such as "gpu:0", other hardware card numbers, such as "npu:0", or CPU, such as "cpu".</td>
+<td><code>gpu:0</code></td>
+</tr>
+<tr>
 <td><code>threshold</code></td>
 <td>Threshold for filtering low-confidence objects</td>
 <td><code>float/None</code></td>
 <td>None</td>
 <td>None</td>
 </tr>
+<tr>
+<td><code>use_hpip</code></td>
+<td>Whether to enable high-performance inference. </td>
+<td><code>bool</code></td>
+<td>None</td>
+<td><code>False</code></td>
+</tr>
 </table>
 
 * The `model_name` must be specified. After specifying `model_name`, the default model parameters built into PaddleX are used. If `model_dir` is specified, the user-defined model is used.

+ 56 - 15
docs/module_usage/tutorials/cv_modules/instance_segmentation.md

@@ -160,21 +160,48 @@ comments: true
 </tr>
 </table>
 
-**测试环境说明:**
-
-- **性能测试环境**
-  - **测试数据集**:<a href="https://cocodataset.org/#home">COCO2017</a>验证集。
-  - **硬件配置**:
-    - GPU:NVIDIA Tesla T4
-    - CPU:Intel Xeon Gold 6271C @ 2.60GHz
-    - 其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **推理模式说明**
-
-| 模式        | GPU配置                          | CPU配置          | 加速技术组合                                |
-|-------------|----------------------------------|------------------|---------------------------------------------|
-| 常规模式    | FP32精度 / 无TRT加速             | FP32精度 / 8线程       | PaddleInference                             |
-| 高性能模式  | 选择先验精度类型和加速策略的最优组合         | FP32精度 / 8线程       | 选择先验最优后端(Paddle/OpenVINO/TRT等) |
+<strong>测试环境说明:</strong>
+
+  <ul>
+      <li><b>性能测试环境</b>
+          <ul>
+          <li><strong>测试数据集:</strong><a href="https://cocodataset.org/#home">COCO2017</a>验证集。 </li>
+              <li><strong>硬件配置:</strong>
+                  <ul>
+                      <li>GPU:NVIDIA Tesla T4</li>
+                      <li>CPU:Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>推理模式说明</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>模式</th>
+            <th>GPU配置</th>
+            <th>CPU配置</th>
+            <th>加速技术组合</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>常规模式</td>
+            <td>FP32精度 / 无TRT加速</td>
+            <td>FP32精度 / 8线程</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>高性能模式</td>
+            <td>选择先验精度类型和加速策略的最优组合</td>
+            <td>FP32精度 / 8线程</td>
+            <td>选择先验最优后端(Paddle/OpenVINO/TRT等)</td>
+        </tr>
+    </tbody>
+</table>
 
 </details>
 
@@ -239,6 +266,20 @@ for res in output:
 <td>无</td>
 </tr>
 <tr>
+<td><code>device</code></td>
+<td>模型推理设备</td>
+<td><code>str</code></td>
+<td>支持指定GPU具体卡号,如“gpu:0”,其他硬件具体卡号,如“npu:0”,CPU如“cpu”。</td>
+<td><code>gpu:0</code></td>
+</tr>
+<tr>
+<td><code>use_hpip</code></td>
+<td>是否启用高性能推理</td>
+<td><code>bool</code></td>
+<td>无</td>
+<td><code>False</code></td>
+</tr>
+<tr>
 <td><code>threshold</code></td>
 <td>低分object过滤阈值</td>
 <td><code>float/None</code></td>

+ 56 - 15
docs/module_usage/tutorials/cv_modules/mainbody_detection.en.md

@@ -31,21 +31,48 @@ Mainbody detection is a fundamental task in object detection, aiming to identify
 </tr>
 </table>
 
-**Test Environment Description**:
-
-- **Performance Test Environment**
-  - **Test Dataset**: PaddleClas mainbody detection dataset.
-  - **Hardware Configuration**:
-    - GPU: NVIDIA Tesla T4
-    - CPU: Intel Xeon Gold 6271C @ 2.60GHz
-    - Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **Inference Mode Description**
-
-| Mode        | GPU Configuration                        | CPU Configuration | Acceleration Technology Combination                   |
-|-------------|----------------------------------------|-------------------|---------------------------------------------------|
-| Normal Mode | FP32 Precision / No TRT Acceleration   | FP32 Precision / 8 Threads | PaddleInference                                 |
-| High-Performance Mode | Optimal combination of pre-selected precision types and acceleration strategies | FP32 Precision / 8 Threads | Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.) |
+<strong>Test Environment Description:</strong>
+
+  <ul>
+      <li><b>Performance Test Environment</b>
+          <ul>
+              <li><strong>Test Dataset:</strong>PaddleClas mainbody detection dataset.</li>
+              <li><strong>Hardware Configuration:</strong>
+                  <ul>
+                      <li>GPU: NVIDIA Tesla T4</li>
+                      <li>CPU: Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>Inference Mode Description</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>Mode</th>
+            <th>GPU Configuration </th>
+            <th>CPU Configuration </th>
+            <th>Acceleration Technology Combination</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>Normal Mode</td>
+            <td>FP32 Precision / No TRT Acceleration</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>High-Performance Mode</td>
+            <td>Optimal combination of pre-selected precision types and acceleration strategies</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.)</td>
+        </tr>
+    </tbody>
+</table>
 
 ## III. Quick Integration  <a id="quick"> </a>
 > ❗ Before quick integration, please install the PaddleX wheel package. For detailed instructions, refer to [PaddleX Local Installation Guide](../../../installation/installation.en.md)
@@ -112,12 +139,26 @@ Related methods, parameters, and explanations are as follows:
 <td>None</td>
 </tr>
 <tr>
+<td><code>device</code></td>
+<td>The device used for model inference</td>
+<td><code>str</code></td>
+<td>It supports specifying specific GPU card numbers, such as "gpu:0", other hardware card numbers, such as "npu:0", or CPU, such as "cpu".</td>
+<td><code>gpu:0</code></td>
+</tr>
+<tr>
 <td><code>threshold</code></td>
 <td>Threshold for filtering low-confidence objects</td>
 <td><code>float/None/dict</code></td>
 <td>None</td>
 <td>None</td>
 </tr>
+<tr>
+<td><code>use_hpip</code></td>
+<td>Whether to enable high-performance inference. </td>
+<td><code>bool</code></td>
+<td>None</td>
+<td><code>False</code></td>
+</tr>
 </table>
 
 * The `model_name` must be specified. After specifying `model_name`, the default model parameters built into PaddleX are used. If `model_dir` is specified, the user-defined model is used.

+ 49 - 15
docs/module_usage/tutorials/cv_modules/mainbody_detection.md

@@ -31,21 +31,48 @@ comments: true
 </tr>
 </table>
 
-**测试环境说明:**
-
-- **性能测试环境**
-  - **测试数据集**:PaddleClas主体检测数据集。
-  - **硬件配置**:
-    - GPU:NVIDIA Tesla T4
-    - CPU:Intel Xeon Gold 6271C @ 2.60GHz
-    - 其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **推理模式说明**
-
-| 模式        | GPU配置                          | CPU配置          | 加速技术组合                                |
-|-------------|----------------------------------|------------------|---------------------------------------------|
-| 常规模式    | FP32精度 / 无TRT加速             | FP32精度 / 8线程       | PaddleInference                             |
-| 高性能模式  | 选择先验精度类型和加速策略的最优组合         | FP32精度 / 8线程       | 选择先验最优后端(Paddle/OpenVINO/TRT等) |
+<strong>测试环境说明:</strong>
+
+  <ul>
+      <li><b>性能测试环境</b>
+          <ul>
+          <li><strong>测试数据集:</strong>PaddleClas主体检测数据集。 </li>
+              <li><strong>硬件配置:</strong>
+                  <ul>
+                      <li>GPU:NVIDIA Tesla T4</li>
+                      <li>CPU:Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>推理模式说明</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>模式</th>
+            <th>GPU配置</th>
+            <th>CPU配置</th>
+            <th>加速技术组合</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>常规模式</td>
+            <td>FP32精度 / 无TRT加速</td>
+            <td>FP32精度 / 8线程</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>高性能模式</td>
+            <td>选择先验精度类型和加速策略的最优组合</td>
+            <td>FP32精度 / 8线程</td>
+            <td>选择先验最优后端(Paddle/OpenVINO/TRT等)</td>
+        </tr>
+    </tbody>
+</table>
 
 ## 三、快速集成
 > ❗ 在快速集成前,请先安装 PaddleX 的 wheel 包,详细请参考 [PaddleX本地安装教程](../../../installation/installation.md)
@@ -108,6 +135,13 @@ for res in output:
 <td>无</td>
 </tr>
 <tr>
+<td><code>device</code></td>
+<td>模型推理设备</td>
+<td><code>str</code></td>
+<td>支持指定GPU具体卡号,如“gpu:0”,其他硬件具体卡号,如“npu:0”,CPU如“cpu”。</td>
+<td><code>gpu:0</code></td>
+</tr>
+<tr>
 <td><code>threshold</code></td>
 <td>低分object过滤阈值</td>
 <td><code>float/None/dict[int, float]</code></td>

+ 56 - 15
docs/module_usage/tutorials/cv_modules/object_detection.en.md

@@ -348,21 +348,48 @@ The object detection module is a crucial component in computer vision systems, r
 </tr>
 </table>
 
-**Test Environment Description**:
-
-- **Performance Test Environment**
-  - **Test Dataset**:  <a href="https://cocodataset.org/#home">COCO2017</a> validation set.
-  - **Hardware Configuration**:
-    - GPU: NVIDIA Tesla T4
-    - CPU: Intel Xeon Gold 6271C @ 2.60GHz
-    - Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **Inference Mode Description**
-
-| Mode        | GPU Configuration                        | CPU Configuration | Acceleration Technology Combination                   |
-|-------------|----------------------------------------|-------------------|---------------------------------------------------|
-| Normal Mode | FP32 Precision / No TRT Acceleration   | FP32 Precision / 8 Threads | PaddleInference                                 |
-| High-Performance Mode | Optimal combination of pre-selected precision types and acceleration strategies | FP32 Precision / 8 Threads | Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.) |
+<strong>Test Environment Description:</strong>
+
+  <ul>
+      <li><b>Performance Test Environment</b>
+          <ul>
+            <li><strong>Test Dataset:</strong> <a href="https://cocodataset.org/#home">COCO2017</a> validation set.</li>
+              <li><strong>Hardware Configuration:</strong>
+                  <ul>
+                      <li>GPU: NVIDIA Tesla T4</li>
+                      <li>CPU: Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>Inference Mode Description</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>Mode</th>
+            <th>GPU Configuration </th>
+            <th>CPU Configuration </th>
+            <th>Acceleration Technology Combination</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>Normal Mode</td>
+            <td>FP32 Precision / No TRT Acceleration</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>High-Performance Mode</td>
+            <td>Optimal combination of pre-selected precision types and acceleration strategies</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.)</td>
+        </tr>
+    </tbody>
+</table>
 
 </details>
 
@@ -432,6 +459,13 @@ Related methods, parameters, and explanations are as follows:
 <td>None</td>
 </tr>
 <tr>
+<td><code>device</code></td>
+<td>The device used for model inference</td>
+<td><code>str</code></td>
+<td>It supports specifying specific GPU card numbers, such as "gpu:0", other hardware card numbers, such as "npu:0", or CPU, such as "cpu".</td>
+<td><code>gpu:0</code></td>
+</tr>
+<tr>
 <td><code>img_size</code></td>
 <td>Size of the input image; if not specified, the default configuration of the PaddleX official model will be used</td>
 <td><code>int/list</code></td>
@@ -450,6 +484,13 @@ Related methods, parameters, and explanations are as follows:
 <td>None</td>
 <td>None</td>
 </tr>
+<tr>
+<td><code>use_hpip</code></td>
+<td>Whether to enable high-performance inference. </td>
+<td><code>bool</code></td>
+<td>None</td>
+<td><code>False</code></td>
+</tr>
 </table>
 
 * The `model_name` must be specified. After specifying `model_name`, the default model parameters built into PaddleX are used. If `model_dir` is specified, the user-defined model is used.

+ 49 - 15
docs/module_usage/tutorials/cv_modules/object_detection.md

@@ -363,21 +363,48 @@ comments: true
 </tr>
 </table>
 
-**测试环境说明:**
-
-- **性能测试环境**
-  - **测试数据集**:<a href="https://cocodataset.org/#home">COCO2017</a>验证集。
-  - **硬件配置**:
-    - GPU:NVIDIA Tesla T4
-    - CPU:Intel Xeon Gold 6271C @ 2.60GHz
-    - 其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **推理模式说明**
-
-| 模式        | GPU配置                          | CPU配置          | 加速技术组合                                |
-|-------------|----------------------------------|------------------|---------------------------------------------|
-| 常规模式    | FP32精度 / 无TRT加速             | FP32精度 / 8线程       | PaddleInference                             |
-| 高性能模式  | 选择先验精度类型和加速策略的最优组合         | FP32精度 / 8线程       | 选择先验最优后端(Paddle/OpenVINO/TRT等) |
+<strong>测试环境说明:</strong>
+
+  <ul>
+      <li><b>性能测试环境</b>
+          <ul>
+            <li><strong>测试数据集:</strong><a href="https://cocodataset.org/#home">COCO2017</a>验证集。 </li>
+              <li><strong>硬件配置:</strong>
+                  <ul>
+                      <li>GPU:NVIDIA Tesla T4</li>
+                      <li>CPU:Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>推理模式说明</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>模式</th>
+            <th>GPU配置</th>
+            <th>CPU配置</th>
+            <th>加速技术组合</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>常规模式</td>
+            <td>FP32精度 / 无TRT加速</td>
+            <td>FP32精度 / 8线程</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>高性能模式</td>
+            <td>选择先验精度类型和加速策略的最优组合</td>
+            <td>FP32精度 / 8线程</td>
+            <td>选择先验最优后端(Paddle/OpenVINO/TRT等)</td>
+        </tr>
+    </tbody>
+</table>
 
 </details>
 
@@ -450,6 +477,13 @@ for res in output:
 <td>无</td>
 </tr>
 <tr>
+<td><code>device</code></td>
+<td>模型推理设备</td>
+<td><code>str</code></td>
+<td>支持指定GPU具体卡号,如“gpu:0”,其他硬件具体卡号,如“npu:0”,CPU如“cpu”。</td>
+<td><code>gpu:0</code></td>
+</tr>
+<tr>
 <td><code>img_size</code></td>
 <td>输入图像大小;如果不指定,将默认使用PaddleX官方模型配置</td>
 <td><code>int/list/None</code></td>

+ 54 - 13
docs/module_usage/tutorials/cv_modules/open_vocabulary_detection.en.md

@@ -39,21 +39,48 @@ Open-vocabulary object detection is an advanced object detection technology aime
 </tr>
 </table>
 
-**Test Environment Description**:
+<strong>Test Environment Description:</strong>
 
-- **Performance Test Environment**
-  - **Test Dataset**: Based on the open vocabulary object detection model trained on the three datasets: O365, GoldG, and Cap4M.
-  - **Hardware Configuration**:
-    - GPU: NVIDIA Tesla T4
-    - CPU: Intel Xeon Gold 6271C @ 2.60GHz
-    - Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
+  <ul>
+      <li><b>Performance Test Environment</b>
+          <ul>
+              <li><strong>Test Dataset:</strong>  Based on the open vocabulary object detection model trained on the three datasets: O365, GoldG, and Cap4M.</li>
+              <li><strong>Hardware Configuration:</strong>
+                  <ul>
+                      <li>GPU: NVIDIA Tesla T4</li>
+                      <li>CPU: Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>Inference Mode Description</b></li>
+  </ul>
 
-- **Inference Mode Description**
-
-| Mode        | GPU Configuration                        | CPU Configuration | Acceleration Technology Combination                   |
-|-------------|----------------------------------------|-------------------|---------------------------------------------------|
-| Normal Mode | FP32 Precision / No TRT Acceleration   | FP32 Precision / 8 Threads | PaddleInference                                 |
-| High-Performance Mode | Optimal combination of pre-selected precision types and acceleration strategies | FP32 Precision / 8 Threads | Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.) |
+<table border="1">
+    <thead>
+        <tr>
+            <th>Mode</th>
+            <th>GPU Configuration </th>
+            <th>CPU Configuration </th>
+            <th>Acceleration Technology Combination</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>Normal Mode</td>
+            <td>FP32 Precision / No TRT Acceleration</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>High-Performance Mode</td>
+            <td>Optimal combination of pre-selected precision types and acceleration strategies</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.)</td>
+        </tr>
+    </tbody>
+</table>
 
 ## III. Quick Integration
 > ❗ Before quick integration, please install the PaddleX wheel package first. For details, please refer to the [PaddleX Local Installation Guide](../../../installation/installation.en.md).
@@ -119,12 +146,26 @@ Related methods, parameters, and explanations are as follows:
 <td>None</td>
 </tr>
 <tr>
+<td><code>device</code></td>
+<td>The device used for model inference</td>
+<td><code>str</code></td>
+<td>It supports specifying specific GPU card numbers, such as "gpu:0", other hardware card numbers, such as "npu:0", or CPU, such as "cpu".</td>
+<td><code>gpu:0</code></td>
+</tr>
+<tr>
 <td><code>thresholds</code></td>
 <td>The filtering thresholds used by the model</td>
 <td><code>dict/None</code></td>
 <td>None</td>
 <td>None</td>
 </tr>
+<tr>
+<td><code>use_hpip</code></td>
+<td>Whether to enable high-performance inference. </td>
+<td><code>bool</code></td>
+<td>None</td>
+<td><code>False</code></td>
+</tr>
 </table>
 
 * The `model_name` must be specified. After specifying `model_name`, the model parameters built into PaddleX will be used by default. If `model_dir` is specified, the user-defined model will be used.

+ 54 - 13
docs/module_usage/tutorials/cv_modules/open_vocabulary_detection.md

@@ -40,21 +40,48 @@ comments: true
 </tr>
 </table>
 
-**测试环境说明:**
+<strong>测试环境说明:</strong>
 
-- **性能测试环境**
-  - **测试数据集**:基于O365,GoldG,Cap4M三个数据集训练的开放词汇目标检测模型。
-  - **硬件配置**:
-    - GPU:NVIDIA Tesla T4
-    - CPU:Intel Xeon Gold 6271C @ 2.60GHz
-    - 其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
+  <ul>
+      <li><b>性能测试环境</b>
+          <ul>
+            <li><strong>测试数据集:</strong>基于O365,GoldG,Cap4M三个数据集训练的开放词汇目标检测模型。 </li>
+              <li><strong>硬件配置:</strong>
+                  <ul>
+                      <li>GPU:NVIDIA Tesla T4</li>
+                      <li>CPU:Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>推理模式说明</b></li>
+  </ul>
 
-- **推理模式说明**
-
-| 模式        | GPU配置                          | CPU配置          | 加速技术组合                                |
-|-------------|----------------------------------|------------------|---------------------------------------------|
-| 常规模式    | FP32精度 / 无TRT加速             | FP32精度 / 8线程       | PaddleInference                             |
-| 高性能模式  | 选择先验精度类型和加速策略的最优组合         | FP32精度 / 8线程       | 选择先验最优后端(Paddle/OpenVINO/TRT等) |
+<table border="1">
+    <thead>
+        <tr>
+            <th>模式</th>
+            <th>GPU配置</th>
+            <th>CPU配置</th>
+            <th>加速技术组合</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>常规模式</td>
+            <td>FP32精度 / 无TRT加速</td>
+            <td>FP32精度 / 8线程</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>高性能模式</td>
+            <td>选择先验精度类型和加速策略的最优组合</td>
+            <td>FP32精度 / 8线程</td>
+            <td>选择先验最优后端(Paddle/OpenVINO/TRT等)</td>
+        </tr>
+    </tbody>
+</table>
 
 
 ## 三、快速集成
@@ -117,6 +144,20 @@ for res in results:
 <td>无</td>
 </tr>
 <tr>
+<td><code>device</code></td>
+<td>模型推理设备</td>
+<td><code>str</code></td>
+<td>支持指定GPU具体卡号,如“gpu:0”,其他硬件具体卡号,如“npu:0”,CPU如“cpu”。</td>
+<td><code>gpu:0</code></td>
+</tr>
+<tr>
+<td><code>use_hpip</code></td>
+<td>是否启用高性能推理</td>
+<td><code>bool</code></td>
+<td>无</td>
+<td><code>False</code></td>
+</tr>
+<tr>
 <td><code>thresholds</code></td>
 <td>模型使用的过滤阈值</td>
 <td><code>dict/None</code></td>

+ 53 - 12
docs/module_usage/tutorials/cv_modules/open_vocabulary_segmentation.en.md

@@ -32,20 +32,47 @@ Open-vocabulary segmentation is an image segmentation task that aims to segment
 </tr>
 </table>
 
-**Test Environment Description**:
+<strong>Test Environment Description:</strong>
 
-- **Performance Test Environment**
-  - **Hardware Configuration**:
-    - GPU: NVIDIA Tesla T4
-    - CPU: Intel Xeon Gold 6271C @ 2.60GHz
-    - Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
+  <ul>
+      <li><b>Performance Test Environment</b>
+          <ul>
+              <li><strong>Hardware Configuration:</strong>
+                  <ul>
+                      <li>GPU: NVIDIA Tesla T4</li>
+                      <li>CPU: Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>Inference Mode Description</b></li>
+  </ul>
 
-- **Inference Mode Description**
-
-| Mode        | GPU Configuration                        | CPU Configuration | Acceleration Technology Combination                   |
-|-------------|----------------------------------------|-------------------|---------------------------------------------------|
-| Normal Mode | FP32 Precision / No TRT Acceleration   | FP32 Precision / 8 Threads | PaddleInference                                 |
-| High-Performance Mode | Optimal combination of pre-selected precision types and acceleration strategies | FP32 Precision / 8 Threads | Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.) |
+<table border="1">
+    <thead>
+        <tr>
+            <th>Mode</th>
+            <th>GPU Configuration </th>
+            <th>CPU Configuration </th>
+            <th>Acceleration Technology Combination</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>Normal Mode</td>
+            <td>FP32 Precision / No TRT Acceleration</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>High-Performance Mode</td>
+            <td>Optimal combination of pre-selected precision types and acceleration strategies</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.)</td>
+        </tr>
+    </tbody>
+</table>
 
 ## III. Quick Integration
 > ❗ Before quick integration, please install the PaddleX wheel package. For details, refer to the [PaddleX Local Installation Guide](../../../installation/installation.en.md).
@@ -118,6 +145,20 @@ Related methods and parameter explanations are as follows:
 <td>None</td>
 <td>None</td>
 </tr>
+<tr>
+<td><code>device</code></td>
+<td>The device used for model inference</td>
+<td><code>str</code></td>
+<td>It supports specifying specific GPU card numbers, such as "gpu:0", other hardware card numbers, such as "npu:0", or CPU, such as "cpu".</td>
+<td><code>gpu:0</code></td>
+</tr>
+<tr>
+<td><code>use_hpip</code></td>
+<td>Whether to enable high-performance inference. </td>
+<td><code>bool</code></td>
+<td>None</td>
+<td><code>False</code></td>
+</tr>
 </table>
 
 * The `model_name` must be specified. After specifying `model_name`, the model parameters built into PaddleX will be used by default. If `model_dir` is specified, the user-defined model will be used.

+ 53 - 12
docs/module_usage/tutorials/cv_modules/open_vocabulary_segmentation.md

@@ -33,20 +33,47 @@ comments: true
 </tr>
 </table>
 
-**测试环境说明:**
+<strong>测试环境说明:</strong>
 
-- **性能测试环境**
-  - **硬件配置**:
-    - GPU:NVIDIA Tesla T4
-    - CPU:Intel Xeon Gold 6271C @ 2.60GHz
-    - 其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
+  <ul>
+      <li><b>性能测试环境</b>
+          <ul>
+              <li><strong>硬件配置:</strong>
+                  <ul>
+                      <li>GPU:NVIDIA Tesla T4</li>
+                      <li>CPU:Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>推理模式说明</b></li>
+  </ul>
 
-- **推理模式说明**
-
-| 模式        | GPU配置                          | CPU配置          | 加速技术组合                                |
-|-------------|----------------------------------|------------------|---------------------------------------------|
-| 常规模式    | FP32精度 / 无TRT加速             | FP32精度 / 8线程       | PaddleInference                             |
-| 高性能模式  | 选择先验精度类型和加速策略的最优组合         | FP32精度 / 8线程       | 选择先验最优后端(Paddle/OpenVINO/TRT等) |
+<table border="1">
+    <thead>
+        <tr>
+            <th>模式</th>
+            <th>GPU配置</th>
+            <th>CPU配置</th>
+            <th>加速技术组合</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>常规模式</td>
+            <td>FP32精度 / 无TRT加速</td>
+            <td>FP32精度 / 8线程</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>高性能模式</td>
+            <td>选择先验精度类型和加速策略的最优组合</td>
+            <td>FP32精度 / 8线程</td>
+            <td>选择先验最优后端(Paddle/OpenVINO/TRT等)</td>
+        </tr>
+    </tbody>
+</table>
 
 
 ## 三、快速集成
@@ -117,6 +144,20 @@ for res in results:
 <td>无</td>
 <td>无</td>
 </tr>
+<tr>
+<td><code>device</code></td>
+<td>模型推理设备</td>
+<td><code>str</code></td>
+<td>支持指定GPU具体卡号,如“gpu:0”,其他硬件具体卡号,如“npu:0”,CPU如“cpu”。</td>
+<td><code>gpu:0</code></td>
+</tr>
+<tr>
+<td><code>use_hpip</code></td>
+<td>是否启用高性能推理</td>
+<td><code>bool</code></td>
+<td>无</td>
+<td><code>False</code></td>
+</tr>
 </table>
 
 * 其中,`model_name` 必须指定,指定 `model_name` 后,默认使用 PaddleX 内置的模型参数,在此基础上,指定 `model_dir` 时,使用用户自定义的模型。

+ 56 - 15
docs/module_usage/tutorials/cv_modules/pedestrian_attribute_recognition.en.md

@@ -33,21 +33,48 @@ Pedestrian attribute recognition is a crucial component in computer vision syste
 </tbody>
 </table>
 
-**Test Environment Description**:
-
-- **Performance Test Environment**
-  - **Test Dataset**: PaddleX's internal self-built dataset.
-  - **Hardware Configuration**:
-    - GPU: NVIDIA Tesla T4
-    - CPU: Intel Xeon Gold 6271C @ 2.60GHz
-    - Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **Inference Mode Description**
-
-| Mode        | GPU Configuration                        | CPU Configuration | Acceleration Technology Combination                   |
-|-------------|----------------------------------------|-------------------|---------------------------------------------------|
-| Normal Mode | FP32 Precision / No TRT Acceleration   | FP32 Precision / 8 Threads | PaddleInference                                 |
-| High-Performance Mode | Optimal combination of pre-selected precision types and acceleration strategies | FP32 Precision / 8 Threads | Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.) |
+<strong>Test Environment Description:</strong>
+
+  <ul>
+      <li><b>Performance Test Environment</b>
+          <ul>
+                <li><strong>Test Dataset:</strong> PaddleX's internal self-built dataset.</li>
+              <li><strong>Hardware Configuration:</strong>
+                  <ul>
+                      <li>GPU: NVIDIA Tesla T4</li>
+                      <li>CPU: Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>Inference Mode Description</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>Mode</th>
+            <th>GPU Configuration </th>
+            <th>CPU Configuration </th>
+            <th>Acceleration Technology Combination</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>Normal Mode</td>
+            <td>FP32 Precision / No TRT Acceleration</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>High-Performance Mode</td>
+            <td>Optimal combination of pre-selected precision types and acceleration strategies</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.)</td>
+        </tr>
+    </tbody>
+</table>
 
 ## <span id="lable">III. Quick Integration</span>
 > ❗ Before quick integration, please install the PaddleX wheel package. For detailed instructions, refer to the [PaddleX Local Installation Guide](../../../installation/installation.en.md)
@@ -129,6 +156,13 @@ Relevant methods, parameters, and explanations are as follows:
 <td>None</td>
 </tr>
 <tr>
+<td><code>device</code></td>
+<td>The device used for model inference</td>
+<td><code>str</code></td>
+<td>It supports specifying specific GPU card numbers, such as "gpu:0", other hardware card numbers, such as "npu:0", or CPU, such as "cpu".</td>
+<td><code>gpu:0</code></td>
+</tr>
+<tr>
 <td><code>threshold</code></td>
 <td>The threshold for vehicle attribute recognition</td>
 <td><code>float/list/dict</code></td>
@@ -138,6 +172,13 @@ Relevant methods, parameters, and explanations are as follows:
 </td>
 <td>0.5</td>
 </tr>
+<tr>
+<td><code>use_hpip</code></td>
+<td>Whether to enable high-performance inference. </td>
+<td><code>bool</code></td>
+<td>None</td>
+<td><code>False</code></td>
+</tr>
 </table>
 
 * The `model_name` must be specified. After specifying `model_name`, PaddleX's built-in model parameters are used by default. If `model_dir` is specified, the user-defined model is used.

+ 49 - 15
docs/module_usage/tutorials/cv_modules/pedestrian_attribute_recognition.md

@@ -33,21 +33,48 @@ comments: true
 </tbody>
 </table>
 
-**测试环境说明:**
-
-- **性能测试环境**
-  - **测试数据集**:PaddleX 内部自建数据集
-  - **硬件配置**:
-    - GPU:NVIDIA Tesla T4
-    - CPU:Intel Xeon Gold 6271C @ 2.60GHz
-    - 其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **推理模式说明**
-
-| 模式        | GPU配置                          | CPU配置          | 加速技术组合                                |
-|-------------|----------------------------------|------------------|---------------------------------------------|
-| 常规模式    | FP32精度 / 无TRT加速             | FP32精度 / 8线程       | PaddleInference                             |
-| 高性能模式  | 选择先验精度类型和加速策略的最优组合         | FP32精度 / 8线程       | 选择先验最优后端(Paddle/OpenVINO/TRT等) |
+<strong>测试环境说明:</strong>
+
+  <ul>
+      <li><b>性能测试环境</b>
+          <ul>
+           <li><strong>测试数据集:</strong>PaddleX 内部自建数据集 </li>
+              <li><strong>硬件配置:</strong>
+                  <ul>
+                      <li>GPU:NVIDIA Tesla T4</li>
+                      <li>CPU:Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>推理模式说明</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>模式</th>
+            <th>GPU配置</th>
+            <th>CPU配置</th>
+            <th>加速技术组合</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>常规模式</td>
+            <td>FP32精度 / 无TRT加速</td>
+            <td>FP32精度 / 8线程</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>高性能模式</td>
+            <td>选择先验精度类型和加速策略的最优组合</td>
+            <td>FP32精度 / 8线程</td>
+            <td>选择先验最优后端(Paddle/OpenVINO/TRT等)</td>
+        </tr>
+    </tbody>
+</table>
 
 ## 三、快速集成
 > ❗ 在快速集成前,请先安装 PaddleX 的 wheel 包,详细请参考 [PaddleX本地安装教程](../../../installation/installation.md)
@@ -128,6 +155,13 @@ for res in output:
 <td>无</td>
 </tr>
 <tr>
+<td><code>device</code></td>
+<td>模型推理设备</td>
+<td><code>str</code></td>
+<td>支持指定GPU具体卡号,如“gpu:0”,其他硬件具体卡号,如“npu:0”,CPU如“cpu”。</td>
+<td><code>gpu:0</code></td>
+</tr>
+<tr>
 <td><code>threshold</code></td>
 <td>行人属性识别阈值</td>
 <td><code>float/list/dict</code></td>

+ 57 - 15
docs/module_usage/tutorials/cv_modules/rotated_object_detection.en.md

@@ -28,21 +28,48 @@ Rotated object detection is a derivative of the object detection module, specifi
 </tr>
 </table>
 
-**Test Environment Description**:
-
-- **Performance Test Environment**
-  - **Test Dataset**: <a href="https://captain-whu.github.io/DOTA/">DOTA</a> validation set.
-  - **Hardware Configuration**:
-    - GPU: NVIDIA Tesla T4
-    - CPU: Intel Xeon Gold 6271C @ 2.60GHz
-    - Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **Inference Mode Description**
-
-| Mode        | GPU Configuration                        | CPU Configuration | Acceleration Technology Combination                   |
-|-------------|----------------------------------------|-------------------|---------------------------------------------------|
-| Normal Mode | FP32 Precision / No TRT Acceleration   | FP32 Precision / 8 Threads | PaddleInference                                 |
-| High-Performance Mode | Optimal combination of pre-selected precision types and acceleration strategies | FP32 Precision / 8 Threads | Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.) |
+<strong>Test Environment Description:</strong>
+
+  <ul>
+      <li><b>Performance Test Environment</b>
+          <ul>
+                 <li><strong>Test Dataset:</strong>  <a href="https://captain-whu.github.io/DOTA/">DOTA</a> validation set.</li>
+              <li><strong>Hardware Configuration:</strong>
+                  <ul>
+                      <li>GPU: NVIDIA Tesla T4</li>
+                      <li>CPU: Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>Inference Mode Description</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>Mode</th>
+            <th>GPU Configuration </th>
+            <th>CPU Configuration </th>
+            <th>Acceleration Technology Combination</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>Normal Mode</td>
+            <td>FP32 Precision / No TRT Acceleration</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>High-Performance Mode</td>
+            <td>Optimal combination of pre-selected precision types and acceleration strategies</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.)</td>
+        </tr>
+    </tbody>
+</table>
 
 
 ## III. Quick Integration
@@ -108,6 +135,20 @@ Related methods and parameter explanations are as follows:
 <td>None</td>
 </tr>
 <tr>
+<td><code>device</code></td>
+<td>The device used for model inference</td>
+<td><code>str</code></td>
+<td>It supports specifying specific GPU card numbers, such as "gpu:0", other hardware card numbers, such as "npu:0", or CPU, such as "cpu".</td>
+<td><code>gpu:0</code></td>
+</tr>
+<tr>
+<td><code>use_hpip</code></td>
+<td>Whether to enable high-performance inference. </td>
+<td><code>bool</code></td>
+<td>None</td>
+<td><code>False</code></td>
+</tr>
+<tr>
 <td><code>threshold</code></td>
 <td>The threshold for filtering low-score objects</td>
 <td><code>float/None/dict</code></td>
@@ -121,6 +162,7 @@ Related methods and parameter explanations are as follows:
 <td>None</td>
 <td>None</td>
 </tr>
+
 </table>
 
 * The `model_name` must be specified. After specifying `model_name`, the model parameters built into PaddleX will be used by default. If `model_dir` is specified, the user-defined model will be used.

+ 56 - 15
docs/module_usage/tutorials/cv_modules/rotated_object_detection.md

@@ -28,21 +28,48 @@ comments: true
 </tr>
 </table>
 
-**测试环境说明:**
-
-- **性能测试环境**
-  - **测试数据集**:<a href="https://captain-whu.github.io/DOTA/">DOTA</a>验证集
-  - **硬件配置**:
-    - GPU:NVIDIA Tesla T4
-    - CPU:Intel Xeon Gold 6271C @ 2.60GHz
-    - 其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **推理模式说明**
-
-| 模式        | GPU配置                          | CPU配置          | 加速技术组合                                |
-|-------------|----------------------------------|------------------|---------------------------------------------|
-| 常规模式    | FP32精度 / 无TRT加速             | FP32精度 / 8线程       | PaddleInference                             |
-| 高性能模式  | 选择先验精度类型和加速策略的最优组合         | FP32精度 / 8线程       | 选择先验最优后端(Paddle/OpenVINO/TRT等) |
+<strong>测试环境说明:</strong>
+
+  <ul>
+      <li><b>性能测试环境</b>
+          <ul>
+           <li><strong>测试数据集:</strong><a href="https://captain-whu.github.io/DOTA/">DOTA</a>验证集</li>
+              <li><strong>硬件配置:</strong>
+                  <ul>
+                      <li>GPU:NVIDIA Tesla T4</li>
+                      <li>CPU:Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>推理模式说明</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>模式</th>
+            <th>GPU配置</th>
+            <th>CPU配置</th>
+            <th>加速技术组合</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>常规模式</td>
+            <td>FP32精度 / 无TRT加速</td>
+            <td>FP32精度 / 8线程</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>高性能模式</td>
+            <td>选择先验精度类型和加速策略的最优组合</td>
+            <td>FP32精度 / 8线程</td>
+            <td>选择先验最优后端(Paddle/OpenVINO/TRT等)</td>
+        </tr>
+    </tbody>
+</table>
 
 ## 三、快速集成
 > ❗ 在快速集成前,请先安装 PaddleX 的 wheel 包,详细请参考 [PaddleX本地安装教程](../../../installation/installation.md)
@@ -103,6 +130,20 @@ for res in output:
 <td>无</td>
 </tr>
 <tr>
+<td><code>device</code></td>
+<td>模型推理设备</td>
+<td><code>str</code></td>
+<td>支持指定GPU具体卡号,如“gpu:0”,其他硬件具体卡号,如“npu:0”,CPU如“cpu”。</td>
+<td><code>gpu:0</code></td>
+</tr>
+<tr>
+<td><code>use_hpip</code></td>
+<td>是否启用高性能推理</td>
+<td><code>bool</code></td>
+<td>无</td>
+<td><code>False</code></td>
+</tr>
+<tr>
 <td><code>threshold</code></td>
 <td>低分object过滤阈值</td>
 <td><code>float/None/dict[int, float]</code></td>

+ 56 - 15
docs/module_usage/tutorials/cv_modules/semantic_segmentation.en.md

@@ -207,21 +207,48 @@ Semantic segmentation is a technique in computer vision that classifies each pix
 </tbody>
 </table>
 
-**Test Environment Description**:
-
-- **Performance Test Environment**
-  - **Test Dataset**: <a href="https://groups.csail.mit.edu/vision/datasets/ADE20K/">ADE20k</a> dataset and <a href="https://www.cityscapes-dataset.com/">Cityscapes</a>dataset.
-  - **Hardware Configuration**:
-    - GPU: NVIDIA Tesla T4
-    - CPU: Intel Xeon Gold 6271C @ 2.60GHz
-    - Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **Inference Mode Description**
-
-| Mode        | GPU Configuration                        | CPU Configuration | Acceleration Technology Combination                   |
-|-------------|----------------------------------------|-------------------|---------------------------------------------------|
-| Normal Mode | FP32 Precision / No TRT Acceleration   | FP32 Precision / 8 Threads | PaddleInference                                 |
-| High-Performance Mode | Optimal combination of pre-selected precision types and acceleration strategies | FP32 Precision / 8 Threads | Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.) |
+<strong>Test Environment Description:</strong>
+
+  <ul>
+      <li><b>Performance Test Environment</b>
+          <ul>
+            <li><strong>Test Dataset:</strong>  <a href="https://groups.csail.mit.edu/vision/datasets/ADE20K/">ADE20k</a> dataset and <a href="https://www.cityscapes-dataset.com/">Cityscapes</a> dataset.</li>
+              <li><strong>Hardware Configuration:</strong>
+                  <ul>
+                      <li>GPU: NVIDIA Tesla T4</li>
+                      <li>CPU: Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>Inference Mode Description</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>Mode</th>
+            <th>GPU Configuration </th>
+            <th>CPU Configuration </th>
+            <th>Acceleration Technology Combination</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>Normal Mode</td>
+            <td>FP32 Precision / No TRT Acceleration</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>High-Performance Mode</td>
+            <td>Optimal combination of pre-selected precision types and acceleration strategies</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.)</td>
+        </tr>
+    </tbody>
+</table>
 
 </details>
 
@@ -286,12 +313,26 @@ Related methods, parameters, and explanations are as follows:
 <td>None</td>
 </tr>
 <tr>
+<td><code>device</code></td>
+<td>The device used for model inference</td>
+<td><code>str</code></td>
+<td>It supports specifying specific GPU card numbers, such as "gpu:0", other hardware card numbers, such as "npu:0", or CPU, such as "cpu".</td>
+<td><code>gpu:0</code></td>
+</tr>
+<tr>
 <td><code>target_size</code></td>
 <td>The resolution used during model prediction</td>
 <td><code>int/tuple</code></td>
 <td><code>None/-1/int/tuple</code></td>
 <td><code>None</code></td>
 </tr>
+<tr>
+<td><code>use_hpip</code></td>
+<td>Whether to enable high-performance inference. </td>
+<td><code>bool</code></td>
+<td>None</td>
+<td><code>False</code></td>
+</tr>
 </table>
 
 * The `model_name` must be specified. After specifying `model_name`, the built-in model parameters of PaddleX are used by default. If `model_dir` is specified, the user-defined model is used.

+ 49 - 15
docs/module_usage/tutorials/cv_modules/semantic_segmentation.md

@@ -208,21 +208,48 @@ comments: true
 </tbody>
 </table>
 
-**测试环境说明:**
-
-- **性能测试环境**
-  - **测试数据集**:<a href="https://groups.csail.mit.edu/vision/datasets/ADE20K/">ADE20k</a>数据集及 <a href="https://www.cityscapes-dataset.com/">Cityscapes</a>数据集。
-  - **硬件配置**:
-    - GPU:NVIDIA Tesla T4
-    - CPU:Intel Xeon Gold 6271C @ 2.60GHz
-    - 其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **推理模式说明**
-
-| 模式        | GPU配置                          | CPU配置          | 加速技术组合                                |
-|-------------|----------------------------------|------------------|---------------------------------------------|
-| 常规模式    | FP32精度 / 无TRT加速             | FP32精度 / 8线程       | PaddleInference                             |
-| 高性能模式  | 选择先验精度类型和加速策略的最优组合         | FP32精度 / 8线程       | 选择先验最优后端(Paddle/OpenVINO/TRT等) |
+<strong>测试环境说明:</strong>
+
+  <ul>
+      <li><b>性能测试环境</b>
+          <ul>
+            <li><strong>测试数据集:</strong><a href="https://groups.csail.mit.edu/vision/datasets/ADE20K">ADE20k</a>数据集及 <a href="https://www.cityscapes-dataset.com/">Cityscapes</a>数据集。</li>
+              <li><strong>硬件配置:</strong>
+                  <ul>
+                      <li>GPU:NVIDIA Tesla T4</li>
+                      <li>CPU:Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>推理模式说明</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>模式</th>
+            <th>GPU配置</th>
+            <th>CPU配置</th>
+            <th>加速技术组合</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>常规模式</td>
+            <td>FP32精度 / 无TRT加速</td>
+            <td>FP32精度 / 8线程</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>高性能模式</td>
+            <td>选择先验精度类型和加速策略的最优组合</td>
+            <td>FP32精度 / 8线程</td>
+            <td>选择先验最优后端(Paddle/OpenVINO/TRT等)</td>
+        </tr>
+    </tbody>
+</table>
 
 </details>
 
@@ -282,6 +309,13 @@ for res in output:
 <td>无</td>
 </tr>
 <tr>
+<td><code>device</code></td>
+<td>模型推理设备</td>
+<td><code>str</code></td>
+<td>支持指定GPU具体卡号,如“gpu:0”,其他硬件具体卡号,如“npu:0”,CPU如“cpu”。</td>
+<td><code>gpu:0</code></td>
+</tr>
+<tr>
 <td><code>target_size</code></td>
 <td>模型预测时分辨率</td>
 <td><code>int/tuple</code></td>

+ 56 - 15
docs/module_usage/tutorials/cv_modules/small_object_detection.en.md

@@ -46,21 +46,48 @@ Small object detection typically refers to accurately detecting and locating sma
 </tr>
 </table>
 
-**Test Environment Description**:
-
-- **Performance Test Environment**
-  - **Test Dataset**: VisDrone-DET dataset.
-  - **Hardware Configuration**:
-    - GPU: NVIDIA Tesla T4
-    - CPU: Intel Xeon Gold 6271C @ 2.60GHz
-    - Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **Inference Mode Description**
-
-| Mode        | GPU Configuration                        | CPU Configuration | Acceleration Technology Combination                   |
-|-------------|----------------------------------------|-------------------|---------------------------------------------------|
-| Normal Mode | FP32 Precision / No TRT Acceleration   | FP32 Precision / 8 Threads | PaddleInference                                 |
-| High-Performance Mode | Optimal combination of pre-selected precision types and acceleration strategies | FP32 Precision / 8 Threads | Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.) |
+<strong>Test Environment Description:</strong>
+
+  <ul>
+      <li><b>Performance Test Environment</b>
+          <ul>
+                <li><strong>Test Dataset:</strong>  VisDrone-DET dataset.</li>
+              <li><strong>Hardware Configuration:</strong>
+                  <ul>
+                      <li>GPU: NVIDIA Tesla T4</li>
+                      <li>CPU: Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>Inference Mode Description</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>Mode</th>
+            <th>GPU Configuration </th>
+            <th>CPU Configuration </th>
+            <th>Acceleration Technology Combination</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>Normal Mode</td>
+            <td>FP32 Precision / No TRT Acceleration</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>High-Performance Mode</td>
+            <td>Optimal combination of pre-selected precision types and acceleration strategies</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.)</td>
+        </tr>
+    </tbody>
+</table>
 
 
 ## III. Quick Integration  <a id="quick"> </a>
@@ -130,12 +157,26 @@ Related methods, parameters, and explanations are as follows:
 <td>None</td>
 </tr>
 <tr>
+<td><code>device</code></td>
+<td>The device used for model inference</td>
+<td><code>str</code></td>
+<td>It supports specifying specific GPU card numbers, such as "gpu:0", other hardware card numbers, such as "npu:0", or CPU, such as "cpu".</td>
+<td><code>gpu:0</code></td>
+</tr>
+<tr>
 <td><code>threshold</code></td>
 <td>Threshold for filtering low-confidence objects</td>
 <td><code>float/None/dict</code></td>
 <td>None</td>
 <td>None</td>
 </tr>
+<tr>
+<td><code>use_hpip</code></td>
+<td>Whether to enable high-performance inference. </td>
+<td><code>bool</code></td>
+<td>None</td>
+<td><code>False</code></td>
+</tr>
 </table>
 
 * The `model_name` must be specified. After specifying `model_name`, the default model parameters built into PaddleX are used. If `model_dir` is specified, the user-defined model is used.

+ 49 - 15
docs/module_usage/tutorials/cv_modules/small_object_detection.md

@@ -47,21 +47,48 @@ comments: true
 </tr>
 </table>
 
-**测试环境说明:**
-
-- **性能测试环境**
-  - **测试数据集**:VisDrone-DET 验证集。
-  - **硬件配置**:
-    - GPU:NVIDIA Tesla T4
-    - CPU:Intel Xeon Gold 6271C @ 2.60GHz
-    - 其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **推理模式说明**
-
-| 模式        | GPU配置                          | CPU配置          | 加速技术组合                                |
-|-------------|----------------------------------|------------------|---------------------------------------------|
-| 常规模式    | FP32精度 / 无TRT加速             | FP32精度 / 8线程       | PaddleInference                             |
-| 高性能模式  | 选择先验精度类型和加速策略的最优组合         | FP32精度 / 8线程       | 选择先验最优后端(Paddle/OpenVINO/TRT等) |
+<strong>测试环境说明:</strong>
+
+  <ul>
+      <li><b>性能测试环境</b>
+          <ul>
+            <li><strong>测试数据集:</strong>VisDrone-DET 验证集。</li>
+              <li><strong>硬件配置:</strong>
+                  <ul>
+                      <li>GPU:NVIDIA Tesla T4</li>
+                      <li>CPU:Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>推理模式说明</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>模式</th>
+            <th>GPU配置</th>
+            <th>CPU配置</th>
+            <th>加速技术组合</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>常规模式</td>
+            <td>FP32精度 / 无TRT加速</td>
+            <td>FP32精度 / 8线程</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>高性能模式</td>
+            <td>选择先验精度类型和加速策略的最优组合</td>
+            <td>FP32精度 / 8线程</td>
+            <td>选择先验最优后端(Paddle/OpenVINO/TRT等)</td>
+        </tr>
+    </tbody>
+</table>
 
 
 ## 三、快速集成
@@ -124,6 +151,13 @@ for res in output:
 <td>无</td>
 </tr>
 <tr>
+<td><code>device</code></td>
+<td>模型推理设备</td>
+<td><code>str</code></td>
+<td>支持指定GPU具体卡号,如“gpu:0”,其他硬件具体卡号,如“npu:0”,CPU如“cpu”。</td>
+<td><code>gpu:0</code></td>
+</tr>
+<tr>
 <td><code>threshold</code></td>
 <td>低分object过滤阈值</td>
 <td><code>float/None/dict[int, float]</code></td>

+ 56 - 15
docs/module_usage/tutorials/cv_modules/vehicle_attribute_recognition.en.md

@@ -33,21 +33,48 @@ Vehicle attribute recognition is a crucial component in computer vision systems.
 </tbody>
 </table>
 
-**Test Environment Description**:
-
-- **Performance Test Environment**
-  - **Test Dataset**: VeRi dataset.
-  - **Hardware Configuration**:
-    - GPU: NVIDIA Tesla T4
-    - CPU: Intel Xeon Gold 6271C @ 2.60GHz
-    - Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **Inference Mode Description**
-
-| Mode        | GPU Configuration                        | CPU Configuration | Acceleration Technology Combination                   |
-|-------------|----------------------------------------|-------------------|---------------------------------------------------|
-| Normal Mode | FP32 Precision / No TRT Acceleration   | FP32 Precision / 8 Threads | PaddleInference                                 |
-| High-Performance Mode | Optimal combination of pre-selected precision types and acceleration strategies | FP32 Precision / 8 Threads | Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.) |
+<strong>Test Environment Description:</strong>
+
+  <ul>
+      <li><b>Performance Test Environment</b>
+          <ul>
+           <li><strong>Test Dataset:</strong>  VeRi dataset.</li>
+              <li><strong>Hardware Configuration:</strong>
+                  <ul>
+                      <li>GPU: NVIDIA Tesla T4</li>
+                      <li>CPU: Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>Inference Mode Description</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>Mode</th>
+            <th>GPU Configuration </th>
+            <th>CPU Configuration </th>
+            <th>Acceleration Technology Combination</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>Normal Mode</td>
+            <td>FP32 Precision / No TRT Acceleration</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>High-Performance Mode</td>
+            <td>Optimal combination of pre-selected precision types and acceleration strategies</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.)</td>
+        </tr>
+    </tbody>
+</table>
 
 
 ## <span id="lable">III. Quick Integration</span>
@@ -111,6 +138,13 @@ Relevant methods, parameters, and explanations are as follows:
 <td>None</td>
 </tr>
 <tr>
+<td><code>device</code></td>
+<td>The device used for model inference</td>
+<td><code>str</code></td>
+<td>It supports specifying specific GPU card numbers, such as "gpu:0", other hardware card numbers, such as "npu:0", or CPU, such as "cpu".</td>
+<td><code>gpu:0</code></td>
+</tr>
+<tr>
 <td><code>threshold</code></td>
 <td>The threshold for vehicle attribute recognition</td>
 <td><code>float/list/dict</code></td>
@@ -120,6 +154,13 @@ Relevant methods, parameters, and explanations are as follows:
 </td>
 <td>0.5</td>
 </tr>
+<tr>
+<td><code>use_hpip</code></td>
+<td>Whether to enable high-performance inference. </td>
+<td><code>bool</code></td>
+<td>None</td>
+<td><code>False</code></td>
+</tr>
 </table>
 
 * The `model_name` must be specified. After specifying `model_name`, PaddleX's built-in model parameters are used by default. If `model_dir` is specified, the user-defined model is used.

+ 49 - 15
docs/module_usage/tutorials/cv_modules/vehicle_attribute_recognition.md

@@ -33,21 +33,48 @@ comments: true
 </tbody>
 </table>
 
-**测试环境说明:**
-
-- **性能测试环境**
-  - **测试数据集**:VeRi 数据集
-  - **硬件配置**:
-    - GPU:NVIDIA Tesla T4
-    - CPU:Intel Xeon Gold 6271C @ 2.60GHz
-    - 其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **推理模式说明**
-
-| 模式        | GPU配置                          | CPU配置          | 加速技术组合                                |
-|-------------|----------------------------------|------------------|---------------------------------------------|
-| 常规模式    | FP32精度 / 无TRT加速             | FP32精度 / 8线程       | PaddleInference                             |
-| 高性能模式  | 选择先验精度类型和加速策略的最优组合         | FP32精度 / 8线程       | 选择先验最优后端(Paddle/OpenVINO/TRT等) |
+<strong>测试环境说明:</strong>
+
+  <ul>
+      <li><b>性能测试环境</b>
+          <ul>
+             <li><strong>测试数据集:</strong>VeRi 数据集</li>
+              <li><strong>硬件配置:</strong>
+                  <ul>
+                      <li>GPU:NVIDIA Tesla T4</li>
+                      <li>CPU:Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>推理模式说明</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>模式</th>
+            <th>GPU配置</th>
+            <th>CPU配置</th>
+            <th>加速技术组合</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>常规模式</td>
+            <td>FP32精度 / 无TRT加速</td>
+            <td>FP32精度 / 8线程</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>高性能模式</td>
+            <td>选择先验精度类型和加速策略的最优组合</td>
+            <td>FP32精度 / 8线程</td>
+            <td>选择先验最优后端(Paddle/OpenVINO/TRT等)</td>
+        </tr>
+    </tbody>
+</table>
 
 
 ## 三、快速集成
@@ -109,6 +136,13 @@ for res in output:
 <td>无</td>
 </tr>
 <tr>
+<td><code>device</code></td>
+<td>模型推理设备</td>
+<td><code>str</code></td>
+<td>支持指定GPU具体卡号,如“gpu:0”,其他硬件具体卡号,如“npu:0”,CPU如“cpu”。</td>
+<td><code>gpu:0</code></td>
+</tr>
+<tr>
 <td><code>threshold</code></td>
 <td>车辆属性识别阈值</td>
 <td><code>float/list/dict</code></td>

+ 56 - 15
docs/module_usage/tutorials/cv_modules/vehicle_detection.en.md

@@ -36,21 +36,48 @@ Vehicle detection is a subtask of object detection, specifically referring to th
 </tr>
 </table>
 
-**Test Environment Description**:
-
-- **Performance Test Environment**
-  - **Test Dataset**: PPVehicle dataset.
-  - **Hardware Configuration**:
-    - GPU: NVIDIA Tesla T4
-    - CPU: Intel Xeon Gold 6271C @ 2.60GHz
-    - Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **Inference Mode Description**
-
-| Mode        | GPU Configuration                        | CPU Configuration | Acceleration Technology Combination                   |
-|-------------|----------------------------------------|-------------------|---------------------------------------------------|
-| Normal Mode | FP32 Precision / No TRT Acceleration   | FP32 Precision / 8 Threads | PaddleInference                                 |
-| High-Performance Mode | Optimal combination of pre-selected precision types and acceleration strategies | FP32 Precision / 8 Threads | Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.) |
+<strong>Test Environment Description:</strong>
+
+  <ul>
+      <li><b>Performance Test Environment</b>
+          <ul>
+           <li><strong>Test Dataset:</strong>  PPVehicle dataset.</li>
+              <li><strong>Hardware Configuration:</strong>
+                  <ul>
+                      <li>GPU: NVIDIA Tesla T4</li>
+                      <li>CPU: Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>Inference Mode Description</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>Mode</th>
+            <th>GPU Configuration </th>
+            <th>CPU Configuration </th>
+            <th>Acceleration Technology Combination</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>Normal Mode</td>
+            <td>FP32 Precision / No TRT Acceleration</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>High-Performance Mode</td>
+            <td>Optimal combination of pre-selected precision types and acceleration strategies</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.)</td>
+        </tr>
+    </tbody>
+</table>
 
 ## III. Quick Integration
 > ❗ Before quick integration, please install the PaddleX wheel package. For detailed instructions, refer to the [PaddleX Local Installation Guide](../../../installation/installation.en.md)
@@ -115,12 +142,26 @@ Related methods, parameters, and explanations are as follows:
 <td><code>None</code></td>
 </tr>
 <tr>
+<td><code>device</code></td>
+<td>The device used for model inference</td>
+<td><code>str</code></td>
+<td>It supports specifying specific GPU card numbers, such as "gpu:0", other hardware card numbers, such as "npu:0", or CPU, such as "cpu".</td>
+<td><code>gpu:0</code></td>
+</tr>
+<tr>
 <td><code>threshold</code></td>
 <td>The threshold for filtering low-score objects</td>
 <td><code>float/None/dict</code></td>
 <td>None</td>
 <td><code>None</code></td>
 </tr>
+<tr>
+<td><code>use_hpip</code></td>
+<td>Whether to enable high-performance inference. </td>
+<td><code>bool</code></td>
+<td>None</td>
+<td><code>False</code></td>
+</tr>
 </table>
 
 * The `model_name` must be specified. After specifying `model_name`, the built-in model parameters of PaddleX are used by default. If `model_dir` is specified, the user-defined model is used.

+ 49 - 15
docs/module_usage/tutorials/cv_modules/vehicle_detection.md

@@ -36,21 +36,48 @@ comments: true
 </tr>
 </table>
 
-**测试环境说明:**
-
-- **性能测试环境**
-  - **测试数据集**:PPVehicle 验证集。
-  - **硬件配置**:
-    - GPU:NVIDIA Tesla T4
-    - CPU:Intel Xeon Gold 6271C @ 2.60GHz
-    - 其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **推理模式说明**
-
-| 模式        | GPU配置                          | CPU配置          | 加速技术组合                                |
-|-------------|----------------------------------|------------------|---------------------------------------------|
-| 常规模式    | FP32精度 / 无TRT加速             | FP32精度 / 8线程       | PaddleInference                             |
-| 高性能模式  | 选择先验精度类型和加速策略的最优组合         | FP32精度 / 8线程       | 选择先验最优后端(Paddle/OpenVINO/TRT等) |
+<strong>测试环境说明:</strong>
+
+  <ul>
+      <li><b>性能测试环境</b>
+          <ul>
+           <li><strong>测试数据集:</strong>PPVehicle 验证集。</li>
+              <li><strong>硬件配置:</strong>
+                  <ul>
+                      <li>GPU:NVIDIA Tesla T4</li>
+                      <li>CPU:Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>推理模式说明</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>模式</th>
+            <th>GPU配置</th>
+            <th>CPU配置</th>
+            <th>加速技术组合</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>常规模式</td>
+            <td>FP32精度 / 无TRT加速</td>
+            <td>FP32精度 / 8线程</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>高性能模式</td>
+            <td>选择先验精度类型和加速策略的最优组合</td>
+            <td>FP32精度 / 8线程</td>
+            <td>选择先验最优后端(Paddle/OpenVINO/TRT等)</td>
+        </tr>
+    </tbody>
+</table>
 
 ## 三、快速集成
 > ❗ 在快速集成前,请先安装 PaddleX 的 wheel 包,详细请参考 [PaddleX本地安装教程](../../../installation/installation.md)
@@ -112,6 +139,13 @@ for res in output:
 <td>无</td>
 </tr>
 <tr>
+<td><code>device</code></td>
+<td>模型推理设备</td>
+<td><code>str</code></td>
+<td>支持指定GPU具体卡号,如“gpu:0”,其他硬件具体卡号,如“npu:0”,CPU如“cpu”。</td>
+<td><code>gpu:0</code></td>
+</tr>
+<tr>
 <td><code>threshold</code></td>
 <td>低分object过滤阈值</td>
 <td><code>float/None/dict[int, float]</code></td>

+ 56 - 15
docs/module_usage/tutorials/ocr_modules/doc_img_orientation_classification.en.md

@@ -33,21 +33,48 @@ The document image orientation classification module is aim to distinguish the o
 </tbody>
 </table>
 
-**Test Environment Description**:
-
-- **Performance Test Environment**
-  - **Test Dataset**: Self-built multi-scene dataset (1000 images, including ID cards/documents, etc.)
-  - **Hardware Configuration**:
-    - GPU: NVIDIA Tesla T4
-    - CPU: Intel Xeon Gold 6271C @ 2.60GHz
-    - Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **Inference Mode Description**
-
-| Mode        | GPU Configuration                        | CPU Configuration | Acceleration Technology Combination                   |
-|-------------|----------------------------------------|-------------------|---------------------------------------------------|
-| Normal Mode | FP32 Precision / No TRT Acceleration   | FP32 Precision / 8 Threads | PaddleInference                                 |
-| High-Performance Mode | Optimal combination of pre-selected precision types and acceleration strategies | FP32 Precision / 8 Threads | Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.) |
+<strong>Test Environment Description:</strong>
+
+  <ul>
+      <li><b>Performance Test Environment</b>
+          <ul>
+             <li><strong>Test Dataset:</strong> Self-built multi-scene dataset (1000 images, including ID cards/documents, etc.)</li>
+              <li><strong>Hardware Configuration:</strong>
+                  <ul>
+                      <li>GPU: NVIDIA Tesla T4</li>
+                      <li>CPU: Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>Inference Mode Description</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>Mode</th>
+            <th>GPU Configuration </th>
+            <th>CPU Configuration </th>
+            <th>Acceleration Technology Combination</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>Normal Mode</td>
+            <td>FP32 Precision / No TRT Acceleration</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>High-Performance Mode</td>
+            <td>Optimal combination of pre-selected precision types and acceleration strategies</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.)</td>
+        </tr>
+    </tbody>
+</table>
 
 ## III. Quick Integration
 
@@ -109,6 +136,20 @@ Related methods, parameters, and other explanations are as follows:
 <td>No</td>
 <td>None</td>
 </tr>
+<tr>
+<td><code>device</code></td>
+<td>The device used for model inference</td>
+<td><code>str</code></td>
+<td>It supports specifying specific GPU card numbers, such as "gpu:0", other hardware card numbers, such as "npu:0", or CPU, such as "cpu".</td>
+<td><code>gpu:0</code></td>
+</tr>
+<tr>
+<td><code>use_hpip</code></td>
+<td>Whether to enable high-performance inference. </td>
+<td><code>bool</code></td>
+<td>None</td>
+<td><code>False</code></td>
+</tr>
 </table>
 
 * The `model_name` must be specified. After specifying `model_name`, the default model parameters built into PaddleX will be used. If `model_dir` is specified, the user-defined model will be used.

+ 49 - 15
docs/module_usage/tutorials/ocr_modules/doc_img_orientation_classification.md

@@ -33,21 +33,48 @@ comments: true
 </tbody>
 </table>
 
-**测试环境说明:**
-
-- **性能测试环境**
-  - **测试数据集**:自建多场景数据集(1000张图片,含证件/文档等场景)
-  - **硬件配置**:
-    - GPU:NVIDIA Tesla T4
-    - CPU:Intel Xeon Gold 6271C @ 2.60GHz
-    - 其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **推理模式说明**
-
-| 模式        | GPU配置                          | CPU配置          | 加速技术组合                                |
-|-------------|----------------------------------|------------------|---------------------------------------------|
-| 常规模式    | FP32精度 / 无TRT加速             | FP32精度 / 8线程       | PaddleInference                             |
-| 高性能模式  | 选择先验精度类型和加速策略的最优组合         | FP32精度 / 8线程       | 选择先验最优后端(Paddle/OpenVINO/TRT等) |
+ <strong>测试环境说明:</strong>
+
+  <ul>
+      <li><b>性能测试环境</b>
+          <ul>
+              <li><strong>测试数据集:</strong>自建多场景数据集(1000张图片,含证件/文档等场景)</li>
+              <li><strong>硬件配置:</strong>
+                  <ul>
+                      <li>GPU:NVIDIA Tesla T4</li>
+                      <li>CPU:Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>推理模式说明</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>模式</th>
+            <th>GPU配置</th>
+            <th>CPU配置</th>
+            <th>加速技术组合</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>常规模式</td>
+            <td>FP32精度 / 无TRT加速</td>
+            <td>FP32精度 / 8线程</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>高性能模式</td>
+            <td>选择先验精度类型和加速策略的最优组合</td>
+            <td>FP32精度 / 8线程</td>
+            <td>选择先验最优后端(Paddle/OpenVINO/TRT等)</td>
+        </tr>
+    </tbody>
+</table>
 
 ## 三、快速集成
 
@@ -109,6 +136,13 @@ for res in output:
 <td>无</td>
 </tr>
 <tr>
+<td><code>device</code></td>
+<td>模型推理设备</td>
+<td><code>str</code></td>
+<td>支持指定GPU具体卡号,如“gpu:0”,其他硬件具体卡号,如“npu:0”,CPU如“cpu”。</td>
+<td><code>gpu:0</code></td>
+</tr>
+<tr>
 <td><code>use_hpip</code></td>
 <td>是否启用高性能推理</td>
 <td><code>bool</code></td>

+ 49 - 15
docs/module_usage/tutorials/ocr_modules/formula_recognition.en.md

@@ -52,21 +52,48 @@ The formula recognition module is a crucial component of OCR (Optical Character
 
 <b>Note: The above accuracy metrics are measured using an internally built formula recognition test set within PaddleX. The BLEU score of LaTeX_OCR_rec on the LaTeX-OCR formula recognition test set is 0.8821. All model GPU inference times are based on machines with Tesla V100 GPUs, with precision type FP32.</b>
 
-**Test Environment Description**:
-
-- **Performance Test Environment**
-  - **Test Dataset**: PaddleX Internal Self-built Formula Recognition Test Set
-  - **Hardware Configuration**:
-    - GPU: NVIDIA Tesla T4
-    - CPU: Intel Xeon Gold 6271C @ 2.60GHz
-    - Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **Inference Mode Description**
-
-| Mode        | GPU Configuration                        | CPU Configuration | Acceleration Technology Combination                   |
-|-------------|----------------------------------------|-------------------|---------------------------------------------------|
-| Normal Mode | FP32 Precision / No TRT Acceleration   | FP32 Precision / 8 Threads | PaddleInference                                 |
-| High-Performance Mode | Optimal combination of pre-selected precision types and acceleration strategies | FP32 Precision / 8 Threads | Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.) |
+<strong>Test Environment Description:</strong>
+
+  <ul>
+      <li><b>Performance Test Environment</b>
+          <ul>
+              <li><strong>Test Dataset:</strong>PaddleX Internal Self-built Formula Recognition Test Set</li>
+              <li><strong>Hardware Configuration:</strong>
+                  <ul>
+                      <li>GPU: NVIDIA Tesla T4</li>
+                      <li>CPU: Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>Inference Mode Description</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>Mode</th>
+            <th>GPU Configuration </th>
+            <th>CPU Configuration </th>
+            <th>Acceleration Technology Combination</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>Normal Mode</td>
+            <td>FP32 Precision / No TRT Acceleration</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>High-Performance Mode</td>
+            <td>Optimal combination of pre-selected precision types and acceleration strategies</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.)</td>
+        </tr>
+    </tbody>
+</table>
 
 ## III. Quick Integration
 > ❗ Before quick integration, please install the PaddleX wheel package. For details, please refer to the [PaddleX Local Installation Guide](../../../installation/installation.en.md)
@@ -132,6 +159,13 @@ The explanations for the methods, parameters, etc., are as follows:
 <td>None</td>
 </tr>
 <tr>
+<td><code>device</code></td>
+<td>The device used for model inference</td>
+<td><code>str</code></td>
+<td>It supports specifying specific GPU card numbers, such as "gpu:0", other hardware card numbers, such as "npu:0", or CPU, such as "cpu".</td>
+<td><code>gpu:0</code></td>
+</tr>
+<tr>
 <td><code>use_hpip</code></td>
 <td>Whether to enable high-performance inference. </td>
 <td><code>bool</code></td>

+ 49 - 16
docs/module_usage/tutorials/ocr_modules/formula_recognition.md

@@ -47,22 +47,48 @@ comments: true
 <td>LaTeX-OCR是一种基于自回归大模型的公式识别算法,通过采用 Hybrid ViT 作为骨干网络,transformer作为解码器,显著提升了公式识别的准确性。</td>
 </tr>
 </table>
-
-**测试环境说明:**
-
-- **性能测试环境**
-  - **测试数据集**:PaddleX 内部自建公式识别测试集
-  - **硬件配置**:
-    - GPU:NVIDIA Tesla T4
-    - CPU:Intel Xeon Gold 6271C @ 2.60GHz
-    - 其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **推理模式说明**
-
-| 模式        | GPU配置                          | CPU配置          | 加速技术组合                                |
-|-------------|----------------------------------|------------------|---------------------------------------------|
-| 常规模式    | FP32精度 / 无TRT加速             | FP32精度 / 8线程       | PaddleInference                             |
-| 高性能模式  | 选择先验精度类型和加速策略的最优组合         | FP32精度 / 8线程       | 选择先验最优后端(Paddle/OpenVINO/TRT等) |
+ <strong>测试环境说明:</strong>
+
+  <ul>
+      <li><b>性能测试环境</b>
+          <ul>
+              <li><strong>测试数据集:</strong>PaddleX 内部自建公式识别测试集</li>
+              <li><strong>硬件配置:</strong>
+                  <ul>
+                      <li>GPU:NVIDIA Tesla T4</li>
+                      <li>CPU:Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>推理模式说明</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>模式</th>
+            <th>GPU配置</th>
+            <th>CPU配置</th>
+            <th>加速技术组合</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>常规模式</td>
+            <td>FP32精度 / 无TRT加速</td>
+            <td>FP32精度 / 8线程</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>高性能模式</td>
+            <td>选择先验精度类型和加速策略的最优组合</td>
+            <td>FP32精度 / 8线程</td>
+            <td>选择先验最优后端(Paddle/OpenVINO/TRT等)</td>
+        </tr>
+    </tbody>
+</table>
 
 
 ## 三、快速集成
@@ -127,6 +153,13 @@ sudo apt-get install texlive texlive-latex-base texlive-latex-extra -y
 <td>无</td>
 </tr>
 <tr>
+<td><code>device</code></td>
+<td>模型推理设备</td>
+<td><code>str</code></td>
+<td>支持指定GPU具体卡号,如“gpu:0”,其他硬件具体卡号,如“npu:0”,CPU如“cpu”。</td>
+<td><code>gpu:0</code></td>
+</tr>
+<tr>
 <td><code>use_hpip</code></td>
 <td>是否启用高性能推理</td>
 <td><code>bool</code></td>

+ 65 - 21
docs/module_usage/tutorials/ocr_modules/layout_detection.en.md

@@ -178,26 +178,56 @@ The core task of structure analysis is to parse and segment the content of input
 </tbody>
 </table>
 
-**Test Environment Description**:
-
-- **Performance Test Environment**
-  - **Test Dataset**:
-    - Layout Detection Model: A self-built layout area detection dataset by PaddleOCR, containing 500 common document type images such as Chinese and English papers, magazines, contracts, books, exam papers, and research reports.
-    - Table Layout Detection Model: A self-built table area detection dataset by PaddleOCR, including 7,835 Chinese and English paper document type images with tables.
-    - 3-Class Layout Detection Model: A self-built layout area detection dataset by PaddleOCR, comprising 1,154 common document type images such as Chinese and English papers, magazines, and research reports.
-    - 5-Class English Document Area Detection Model: The evaluation dataset of [PubLayNet](https://developer.ibm.com/exchanges/data/all/publaynet), containing 11,245 images of English documents.
-    - 17-Class Area Detection Model: A self-built layout area detection dataset by PaddleOCR, including 892 common document type images such as Chinese and English papers, magazines, and research reports.
-  - **Hardware Configuration**:
-    - GPU: NVIDIA Tesla T4
-    - CPU: Intel Xeon Gold 6271C @ 2.60GHz
-    - Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **Inference Mode Description**
-
-| Mode        | GPU Configuration                        | CPU Configuration | Acceleration Technology Combination                   |
-|-------------|----------------------------------------|-------------------|---------------------------------------------------|
-| Normal Mode | FP32 Precision / No TRT Acceleration   | FP32 Precision / 8 Threads | PaddleInference                                 |
-| High-Performance Mode | Optimal combination of pre-selected precision types and acceleration strategies | FP32 Precision / 8 Threads | Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.) |
+<strong>Test Environment Description:</strong>
+
+  <ul>
+      <li><b>Performance Test Environment</b>
+          <ul>
+             <li><strong>Test Dataset:</strong>
+                 <ul>
+                   <li>Layout Detection Model: A self-built layout area detection dataset by PaddleOCR, containing 500 common document type images such as Chinese and English papers, magazines, contracts, books, exam papers, and research reports.</li>
+                   <li>Table Layout Detection Model: A self-built table area detection dataset by PaddleOCR, including 7,835 Chinese and English paper document type images with tables.</li>
+                   <li> 3-Class Layout Detection Model: A self-built layout area detection dataset by PaddleOCR, comprising 1,154 common document type images such as Chinese and English papers, magazines, and research reports.</li>
+                   <li>5-Class English Document Area Detection Model: The evaluation dataset of <a href="https://developer.ibm.com/exchanges/data/all/publaynet">PubLayNet</a>, containing 11,245 images of English documents.</li>
+                   <li>17-Class Area Detection Model: A self-built layout area detection dataset by PaddleOCR, including 892 common document type images such as Chinese and English papers, magazines, and research reports.</li>
+                 </ul>
+             </li>
+              <li><strong>Hardware Configuration:</strong>
+                  <ul>
+                      <li>GPU: NVIDIA Tesla T4</li>
+                      <li>CPU: Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>Inference Mode Description</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>Mode</th>
+            <th>GPU Configuration </th>
+            <th>CPU Configuration </th>
+            <th>Acceleration Technology Combination</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>Normal Mode</td>
+            <td>FP32 Precision / No TRT Acceleration</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>High-Performance Mode</td>
+            <td>Optimal combination of pre-selected precision types and acceleration strategies</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.)</td>
+        </tr>
+    </tbody>
+</table>
 
 </details>
 
@@ -268,6 +298,13 @@ Relevant methods, parameters, and explanations are as follows:
 <td>None</td>
 </tr>
 <tr>
+<td><code>device</code></td>
+<td>The device used for model inference</td>
+<td><code>str</code></td>
+<td>It supports specifying specific GPU card numbers, such as "gpu:0", other hardware card numbers, such as "npu:0", or CPU, such as "cpu".</td>
+<td><code>gpu:0</code></td>
+</tr>
+<tr>
 <td><code>img_size</code></td>
 <td>Size of the input image; if not specified, the default PaddleX official model configuration will be used</td>
 <td><code>int/list/None</code></td>
@@ -332,7 +369,14 @@ Relevant methods, parameters, and explanations are as follows:
 </td>
 <td>None</td>
 </tr>
-</tr></table>
+<tr>
+<td><code>use_hpip</code></td>
+<td>Whether to enable high-performance inference. </td>
+<td><code>bool</code></td>
+<td>None</td>
+<td><code>False</code></td>
+</tr>
+</table>
 
 * Note that `model_name` must be specified. After specifying `model_name`, the default PaddleX built-in model parameters will be used. If `model_dir` is specified, the user-defined model will be used.
 

+ 59 - 22
docs/module_usage/tutorials/ocr_modules/layout_detection.md

@@ -179,26 +179,56 @@ comments: true
 </tbody>
 </table>
 
-**测试环境说明:**
-
-- **性能测试环境**
-  - **测试数据集**:
-    - 版面检测模型: PaddleOCR 自建的版面区域检测数据集,包含中英文论文、杂志、合同、书本、试卷和研报等常见的 500 张文档类型图片。
-    - 表格版面检测模型:PaddleOCR 自建的版面表格区域检测数据集,包含中英文 7835 张带有表格的论文文档类型图片。
-    - 3类版面检测模型:PaddleOCR 自建的版面区域检测数据集,包含中英文论文、杂志和研报等常见的 1154 张文档类型图片。
-    - 5类英文文档区域检测模型:[PubLayNet](https://developer.ibm.com/exchanges/data/all/publaynet) 的评估数据集,包含英文文档的 11245 张文图片。
-    - 17类区域检测模型:PaddleOCR 自建的版面区域检测数据集,包含中英文论文、杂志和研报等常见的 892 张文档类型图片。
-  - **硬件配置**:
-    - GPU:NVIDIA Tesla T4
-    - CPU:Intel Xeon Gold 6271C @ 2.60GHz
-    - 其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **推理模式说明**
-
-| 模式        | GPU配置                          | CPU配置          | 加速技术组合                                |
-|-------------|----------------------------------|------------------|---------------------------------------------|
-| 常规模式    | FP32精度 / 无TRT加速             | FP32精度 / 8线程       | PaddleInference                             |
-| 高性能模式  | 选择先验精度类型和加速策略的最优组合         | FP32精度 / 8线程       | 选择先验最优后端(Paddle/OpenVINO/TRT等) |
+ <strong>测试环境说明:</strong>
+
+  <ul>
+      <li><b>性能测试环境</b>
+          <ul>
+              <li><strong>测试数据集:</strong>
+                 <ul>
+                    <li>版面检测模型: PaddleOCR 自建的版面区域检测数据集,包含中英文论文、杂志、合同、书本、试卷和研报等常见的 500 张文档类型图片。</li>
+                    <li>表格版面检测模型:PaddleOCR 自建的版面表格区域检测数据集,包含中英文 7835 张带有表格的论文文档类型图片。</li>
+                    <li>3类版面检测模型:PaddleOCR 自建的版面区域检测数据集,包含中英文论文、杂志和研报等常见的 1154 张文档类型图片。</li>
+                    <li>5类英文文档区域检测模型: <a href="https://developer.ibm.com/exchanges/data/all/publaynet" target="_blank">PubLayNet</a> 的评估数据集,包含英文文档的 11245 张文图片。</li>
+                    <li>17类区域检测模型:PaddleOCR 自建的版面区域检测数据集,包含中英文论文、杂志和研报等常见的 892 张文档类型图片。</li>
+                 </ul>
+              </li>
+              <li><strong>硬件配置:</strong>
+                  <ul>
+                      <li>GPU:NVIDIA Tesla T4</li>
+                      <li>CPU:Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>推理模式说明</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>模式</th>
+            <th>GPU配置</th>
+            <th>CPU配置</th>
+            <th>加速技术组合</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>常规模式</td>
+            <td>FP32精度 / 无TRT加速</td>
+            <td>FP32精度 / 8线程</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>高性能模式</td>
+            <td>选择先验精度类型和加速策略的最优组合</td>
+            <td>FP32精度 / 8线程</td>
+            <td>选择先验最优后端(Paddle/OpenVINO/TRT等)</td>
+        </tr>
+    </tbody>
+</table>
 </details>
 
 ## 三、快速集成
@@ -268,6 +298,13 @@ for res in output:
 <td>无</td>
 </tr>
 <tr>
+<td><code>device</code></td>
+<td>模型推理设备</td>
+<td><code>str</code></td>
+<td>支持指定GPU具体卡号,如“gpu:0”,其他硬件具体卡号,如“npu:0”,CPU如“cpu”。</td>
+<td><code>gpu:0</code></td>
+</tr>
+<tr>
 <td><code>img_size</code></td>
 <td>输入图像大小;如果不指定,将默认使用PaddleX官方模型配置</td>
 <td><code>int/list/None</code></td>
@@ -326,7 +363,7 @@ for res in output:
 <li><b>large</b>, 设置为large时,表示在模型输出的检测框中,对于互相重叠包含的检测框,只保留外部最大的框,删除重叠的内部框。</li>
 <li><b>small</b>, 设置为small,表示在模型输出的检测框中,对于互相重叠包含的检测框,只保留内部被包含的小框,删除重叠的外部框。</li>
 <li><b>union</b>, 不进行框的过滤处理,内外框都保留</li>
-<li><b>dict</b>, 字典的key为<b>int</b>类型,代表<code>cls_id</code>, value为<b>str</b>类型, 如<code>{0: "large", 2: "small"}</code>, 表示对第0类别检测框使用large模式,对第2类别检测框使用small模式</li>  
+<li><b>dict</b>, 字典的key为<b>int</b>类型,代表<code>cls_id</code>, value为<b>str</b>类型, 如<code>{0: "large", 2: "small"}</code>, 表示对第0类别检测框使用large模式,对第2类别检测框使用small模式</li>
 <li><b>None</b>, 不指定,将默认使用PaddleX官方模型配置</li>
 </ul>
 </td>
@@ -422,7 +459,7 @@ for res in output:
 <li><b>large</b>, 设置为large时,表示在模型输出的检测框中,对于互相重叠包含的检测框,只保留外部最大的框,删除重叠的内部框。</li>
 <li><b>small</b>, 设置为small,表示在模型输出的检测框中,对于互相重叠包含的检测框,只保留内部被包含的小框,删除重叠的外部框。</li>
 <li><b>union</b>, 不进行框的过滤处理,内外框都保留</li>
-<li><b>dict</b>, 字典的key为<b>int</b>类型,代表<code>cls_id</code>, value为<b>str</b>类型, 如<code>{0: "large", 2: "small"}</code>, 表示对第0类别检测框使用large模式,对第2类别检测框使用small模式</li>  
+<li><b>dict</b>, 字典的key为<b>int</b>类型,代表<code>cls_id</code>, value为<b>str</b>类型, 如<code>{0: "large", 2: "small"}</code>, 表示对第0类别检测框使用large模式,对第2类别检测框使用small模式</li>
 <li><b>None</b>, 不指定,将默认使用 <code>creat_model</code> 指定的 <code>layout_merge_bboxes_mode</code> 参数,如果 <code>creat_model</code> 也没有指定,则默认使用PaddleX官方模型配置</li>
 </ul>
 </td>

+ 56 - 15
docs/module_usage/tutorials/ocr_modules/seal_text_detection.en.md

@@ -41,21 +41,48 @@ The seal text detection module typically outputs multi-point bounding boxes arou
 </tbody>
 </table>
 
-**Test Environment Description**:
-
-- **Performance Test Environment**
-  - **Test Dataset**: PaddleX Custom Dataset, Containing 500 Images of Circular Stamps.
-  - **Hardware Configuration**:
-    - GPU: NVIDIA Tesla T4
-    - CPU: Intel Xeon Gold 6271C @ 2.60GHz
-    - Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **Inference Mode Description**
-
-| Mode        | GPU Configuration                        | CPU Configuration | Acceleration Technology Combination                   |
-|-------------|----------------------------------------|-------------------|---------------------------------------------------|
-| Normal Mode | FP32 Precision / No TRT Acceleration   | FP32 Precision / 8 Threads | PaddleInference                                 |
-| High-Performance Mode | Optimal combination of pre-selected precision types and acceleration strategies | FP32 Precision / 8 Threads | Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.) |
+<strong>Test Environment Description:</strong>
+
+  <ul>
+      <li><b>Performance Test Environment</b>
+          <ul>
+               <li><strong>Test Dataset:</strong> PaddleX Custom Dataset, Containing 500 Images of Circular Stamps.</li>
+              <li><strong>Hardware Configuration:</strong>
+                  <ul>
+                      <li>GPU: NVIDIA Tesla T4</li>
+                      <li>CPU: Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>Inference Mode Description</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>Mode</th>
+            <th>GPU Configuration </th>
+            <th>CPU Configuration </th>
+            <th>Acceleration Technology Combination</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>Normal Mode</td>
+            <td>FP32 Precision / No TRT Acceleration</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>High-Performance Mode</td>
+            <td>Optimal combination of pre-selected precision types and acceleration strategies</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.)</td>
+        </tr>
+    </tbody>
+</table>
 
 
 ## III. Quick Integration
@@ -117,6 +144,13 @@ The explanations of related methods and parameters are as follows:
 <td>None</td>
 </tr>
 <tr>
+<td><code>device</code></td>
+<td>The device used for model inference</td>
+<td><code>str</code></td>
+<td>It supports specifying specific GPU card numbers, such as "gpu:0", other hardware card numbers, such as "npu:0", or CPU, such as "cpu".</td>
+<td><code>gpu:0</code></td>
+</tr>
+<tr>
 <td><code>limit_side_len</code></td>
 <td>Limit on the side length of the image for detection</td>
 <td><code>int/None</code></td>
@@ -190,6 +224,13 @@ The explanations of related methods and parameters are as follows:
 <td>True/False/None</td>
 <td>None</td>
 </tr>
+<tr>
+<td><code>use_hpip</code></td>
+<td>Whether to enable high-performance inference. </td>
+<td><code>bool</code></td>
+<td>None</td>
+<td><code>False</code></td>
+</tr>
 </table>
 
 * The `model_name` must be specified. After specifying `model_name`, the built-in model parameters of PaddleX will be used by default. On this basis, if `model_dir` is specified, the user-defined model will be used.

+ 49 - 16
docs/module_usage/tutorials/ocr_modules/seal_text_detection.md

@@ -40,22 +40,48 @@ comments: true
 </tr>
 </tbody>
 </table>
-
-**测试环境说明:**
-
-- **性能测试环境**
-  - **测试数据集**:PaddleX自建数据集,包含500张圆形印章图像。
-  - **硬件配置**:
-    - GPU:NVIDIA Tesla T4
-    - CPU:Intel Xeon Gold 6271C @ 2.60GHz
-    - 其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **推理模式说明**
-
-| 模式        | GPU配置                          | CPU配置          | 加速技术组合                                |
-|-------------|----------------------------------|------------------|---------------------------------------------|
-| 常规模式    | FP32精度 / 无TRT加速             | FP32精度 / 8线程       | PaddleInference                             |
-| 高性能模式  | 选择先验精度类型和加速策略的最优组合         | FP32精度 / 8线程       | 选择先验最优后端(Paddle/OpenVINO/TRT等) |
+<strong>测试环境说明:</strong>
+
+  <ul>
+      <li><b>性能测试环境</b>
+          <ul>
+              <li><strong>测试数据集:</strong>PaddleX自建数据集,包含500张圆形印章图像。</li>
+              <li><strong>硬件配置:</strong>
+                  <ul>
+                      <li>GPU:NVIDIA Tesla T4</li>
+                      <li>CPU:Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>推理模式说明</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>模式</th>
+            <th>GPU配置</th>
+            <th>CPU配置</th>
+            <th>加速技术组合</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>常规模式</td>
+            <td>FP32精度 / 无TRT加速</td>
+            <td>FP32精度 / 8线程</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>高性能模式</td>
+            <td>选择先验精度类型和加速策略的最优组合</td>
+            <td>FP32精度 / 8线程</td>
+            <td>选择先验最优后端(Paddle/OpenVINO/TRT等)</td>
+        </tr>
+    </tbody>
+</table>
 
 
 ## 三、快速集成
@@ -117,6 +143,13 @@ for res in output:
 <td>无</td>
 </tr>
 <tr>
+<td><code>device</code></td>
+<td>模型推理设备</td>
+<td><code>str</code></td>
+<td>支持指定GPU具体卡号,如“gpu:0”,其他硬件具体卡号,如“npu:0”,CPU如“cpu”。</td>
+<td><code>gpu:0</code></td>
+</tr>
+<tr>
 <td><code>limit_side_len</code></td>
 <td>检测的图像边长限制</td>
 <td><code>int/None</code></td>

+ 56 - 15
docs/module_usage/tutorials/ocr_modules/table_cells_detection.en.md

@@ -34,21 +34,48 @@ The table cell detection module is a key component of table recognition tasks, r
 </tr>
 </table>
 
-**Test Environment Description**:
-
-- **Performance Test Environment**
-  - **Test Dataset**: PaddleX Internal Self-built Evaluation Dataset.
-  - **Hardware Configuration**:
-    - GPU: NVIDIA Tesla T4
-    - CPU: Intel Xeon Gold 6271C @ 2.60GHz
-    - Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **Inference Mode Description**
-
-| Mode        | GPU Configuration                        | CPU Configuration | Acceleration Technology Combination                   |
-|-------------|----------------------------------------|-------------------|---------------------------------------------------|
-| Normal Mode | FP32 Precision / No TRT Acceleration   | FP32 Precision / 8 Threads | PaddleInference                                 |
-| High-Performance Mode | Optimal combination of pre-selected precision types and acceleration strategies | FP32 Precision / 8 Threads | Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.) |
+<strong>Test Environment Description:</strong>
+
+  <ul>
+      <li><b>Performance Test Environment</b>
+          <ul>
+             <li><strong>Test Dataset:</strong> PaddleX Internal Self-built Evaluation Dataset.</li>
+              <li><strong>Hardware Configuration:</strong>
+                  <ul>
+                      <li>GPU: NVIDIA Tesla T4</li>
+                      <li>CPU: Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>Inference Mode Description</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>Mode</th>
+            <th>GPU Configuration </th>
+            <th>CPU Configuration </th>
+            <th>Acceleration Technology Combination</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>Normal Mode</td>
+            <td>FP32 Precision / No TRT Acceleration</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>High-Performance Mode</td>
+            <td>Optimal combination of pre-selected precision types and acceleration strategies</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.)</td>
+        </tr>
+    </tbody>
+</table>
 
 ## III. Quick Integration
 > ❗ Before quick integration, please install the PaddleX wheel package first. For details, refer to the [PaddleX Local Installation Guide](../../../installation/installation.en.md).
@@ -116,6 +143,20 @@ The following is the explanation of the methods, parameters, etc.:
 <td>None</td>
 </tr>
 <tr>
+<td><code>device</code></td>
+<td>The device used for model inference</td>
+<td><code>str</code></td>
+<td>It supports specifying specific GPU card numbers, such as "gpu:0", other hardware card numbers, such as "npu:0", or CPU, such as "cpu".</td>
+<td><code>gpu:0</code></td>
+</tr>
+<tr>
+<td><code>use_hpip</code></td>
+<td>Whether to enable high-performance inference. </td>
+<td><code>bool</code></td>
+<td>None</td>
+<td><code>False</code></td>
+</tr>
+<tr>
 <td><code>img_size</code></td>
 <td>Size of the input image; if not specified, the default configuration of the PaddleX official model will be used</td>
 <td><code>int/list</code></td>

+ 56 - 15
docs/module_usage/tutorials/ocr_modules/table_cells_detection.md

@@ -34,21 +34,48 @@ comments: true
 </tr>
 </table>
 
-**测试环境说明:**
-
-- **性能测试环境**
-  - **测试数据集**:PaddleX 内部自建评测集。
-  - **硬件配置**:
-    - GPU:NVIDIA Tesla T4
-    - CPU:Intel Xeon Gold 6271C @ 2.60GHz
-    - 其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **推理模式说明**
-
-| 模式        | GPU配置                          | CPU配置          | 加速技术组合                                |
-|-------------|----------------------------------|------------------|---------------------------------------------|
-| 常规模式    | FP32精度 / 无TRT加速             | FP32精度 / 8线程       | PaddleInference                             |
-| 高性能模式  | 选择先验精度类型和加速策略的最优组合         | FP32精度 / 8线程       | 选择先验最优后端(Paddle/OpenVINO/TRT等) |
+<strong>测试环境说明:</strong>
+
+  <ul>
+      <li><b>性能测试环境</b>
+          <ul>
+              <li><strong>测试数据集:</strong>PaddleX 内部自建评测集。</li>
+              <li><strong>硬件配置:</strong>
+                  <ul>
+                      <li>GPU:NVIDIA Tesla T4</li>
+                      <li>CPU:Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>推理模式说明</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>模式</th>
+            <th>GPU配置</th>
+            <th>CPU配置</th>
+            <th>加速技术组合</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>常规模式</td>
+            <td>FP32精度 / 无TRT加速</td>
+            <td>FP32精度 / 8线程</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>高性能模式</td>
+            <td>选择先验精度类型和加速策略的最优组合</td>
+            <td>FP32精度 / 8线程</td>
+            <td>选择先验最优后端(Paddle/OpenVINO/TRT等)</td>
+        </tr>
+    </tbody>
+</table>
 
 ## 三、快速集成
 > ❗ 在快速集成前,请先安装 PaddleX 的 wheel 包,详细请参考 [PaddleX本地安装教程](../../../installation/installation.md)
@@ -115,6 +142,20 @@ for res in output:
 <td>无</td>
 </tr>
 <tr>
+<td><code>device</code></td>
+<td>模型推理设备</td>
+<td><code>str</code></td>
+<td>支持指定GPU具体卡号,如“gpu:0”,其他硬件具体卡号,如“npu:0”,CPU如“cpu”。</td>
+<td><code>gpu:0</code></td>
+</tr>
+<tr>
+<td><code>use_hpip</code></td>
+<td>是否启用高性能推理</td>
+<td><code>bool</code></td>
+<td>无</td>
+<td><code>False</code></td>
+</tr>
+<tr>
 <td><code>img_size</code></td>
 <td>输入图像大小;如果不指定,将默认使用PaddleX官方模型配置</td>
 <td><code>int/list</code></td>

+ 56 - 15
docs/module_usage/tutorials/ocr_modules/table_classification.en.md

@@ -26,21 +26,48 @@ The table classification module is a key component of a computer vision system,
 </tr>
 </table>
 
-**Test Environment Description**:
-
-- **Performance Test Environment**
-  - **Test Dataset**: PaddleX Internal Self-built Evaluation Dataset.
-  - **Hardware Configuration**:
-    - GPU: NVIDIA Tesla T4
-    - CPU: Intel Xeon Gold 6271C @ 2.60GHz
-    - Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **Inference Mode Description**
-
-| Mode        | GPU Configuration                        | CPU Configuration | Acceleration Technology Combination                   |
-|-------------|----------------------------------------|-------------------|---------------------------------------------------|
-| Normal Mode | FP32 Precision / No TRT Acceleration   | FP32 Precision / 8 Threads | PaddleInference                                 |
-| High-Performance Mode | Optimal combination of pre-selected precision types and acceleration strategies | FP32 Precision / 8 Threads | Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.) |
+<strong>Test Environment Description:</strong>
+
+  <ul>
+      <li><b>Performance Test Environment</b>
+          <ul>
+           <li><strong>Test Dataset:</strong> PaddleX Internal Self-built Evaluation Dataset.</li>
+              <li><strong>Hardware Configuration:</strong>
+                  <ul>
+                      <li>GPU: NVIDIA Tesla T4</li>
+                      <li>CPU: Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>Inference Mode Description</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>Mode</th>
+            <th>GPU Configuration </th>
+            <th>CPU Configuration </th>
+            <th>Acceleration Technology Combination</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>Normal Mode</td>
+            <td>FP32 Precision / No TRT Acceleration</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>High-Performance Mode</td>
+            <td>Optimal combination of pre-selected precision types and acceleration strategies</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.)</td>
+        </tr>
+    </tbody>
+</table>
 
 ## III. Quick Integration
 > ❗ Before quick integration, please install the PaddleX wheel package first. For details, please refer to the [PaddleX Local Installation Guide](../../../installation/installation.en.md).
@@ -98,6 +125,20 @@ The descriptions of the related methods and parameters are as follows:
 <td>No</td>
 <td><code>None</code></td>
 </tr>
+<tr>
+<td><code>device</code></td>
+<td>The device used for model inference</td>
+<td><code>str</code></td>
+<td>It supports specifying specific GPU card numbers, such as "gpu:0", other hardware card numbers, such as "npu:0", or CPU, such as "cpu".</td>
+<td><code>gpu:0</code></td>
+</tr>
+<tr>
+<td><code>use_hpip</code></td>
+<td>Whether to enable high-performance inference. </td>
+<td><code>bool</code></td>
+<td>None</td>
+<td><code>False</code></td>
+</tr>
 </table>
 
 * The `model_name` must be specified. After specifying the `model_name`, the default model parameters in PaddleX will be used. On this basis, if `model_dir` is specified, the user-defined model will be used.

+ 56 - 15
docs/module_usage/tutorials/ocr_modules/table_classification.md

@@ -27,21 +27,48 @@ comments: true
 </tr>
 </table>
 
-**测试环境说明:**
-
-- **性能测试环境**
-  - **测试数据集**:PaddleX 内部自建评测数据集。
-  - **硬件配置**:
-    - GPU:NVIDIA Tesla T4
-    - CPU:Intel Xeon Gold 6271C @ 2.60GHz
-    - 其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **推理模式说明**
-
-| 模式        | GPU配置                          | CPU配置          | 加速技术组合                                |
-|-------------|----------------------------------|------------------|---------------------------------------------|
-| 常规模式    | FP32精度 / 无TRT加速             | FP32精度 / 8线程       | PaddleInference                             |
-| 高性能模式  | 选择先验精度类型和加速策略的最优组合         | FP32精度 / 8线程       | 选择先验最优后端(Paddle/OpenVINO/TRT等) |
+<strong>测试环境说明:</strong>
+
+  <ul>
+      <li><b>性能测试环境</b>
+          <ul>
+              <li><strong>测试数据集:</strong>PaddleX 内部自建评测数据集。</li>
+              <li><strong>硬件配置:</strong>
+                  <ul>
+                      <li>GPU:NVIDIA Tesla T4</li>
+                      <li>CPU:Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>推理模式说明</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>模式</th>
+            <th>GPU配置</th>
+            <th>CPU配置</th>
+            <th>加速技术组合</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>常规模式</td>
+            <td>FP32精度 / 无TRT加速</td>
+            <td>FP32精度 / 8线程</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>高性能模式</td>
+            <td>选择先验精度类型和加速策略的最优组合</td>
+            <td>FP32精度 / 8线程</td>
+            <td>选择先验最优后端(Paddle/OpenVINO/TRT等)</td>
+        </tr>
+    </tbody>
+</table>
 
 ## 三、快速集成
 > ❗ 在快速集成前,请先安装 PaddleX 的 wheel 包,详细请参考 [PaddleX本地安装教程](../../../installation/installation.md)。
@@ -100,6 +127,20 @@ for res in output:
 <td>无</td>
 <td>无</td>
 </tr>
+<tr>
+<td><code>device</code></td>
+<td>模型推理设备</td>
+<td><code>str</code></td>
+<td>支持指定GPU具体卡号,如“gpu:0”,其他硬件具体卡号,如“npu:0”,CPU如“cpu”。</td>
+<td><code>gpu:0</code></td>
+</tr>
+<tr>
+<td><code>use_hpip</code></td>
+<td>是否启用高性能推理</td>
+<td><code>bool</code></td>
+<td>无</td>
+<td><code>False</code></td>
+</tr>
 </table>
 
 * 其中,`model_name` 必须指定,指定 `model_name` 后,默认使用 PaddleX 内置的模型参数,在此基础上,指定 `model_dir` 时,使用用户自定义的模型。

+ 56 - 15
docs/module_usage/tutorials/ocr_modules/table_structure_recognition.en.md

@@ -52,21 +52,48 @@ SLANet_plus is an enhanced version of SLANet, a table structure recognition mode
 </tr>
 </table>
 
-**Test Environment Description**:
-
-- **Performance Test Environment**
-  - **Test Dataset**: PaddleX Internal Self-built High-difficulty Chinese Table Recognition Dataset.
-  - **Hardware Configuration**:
-    - GPU: NVIDIA Tesla T4
-    - CPU: Intel Xeon Gold 6271C @ 2.60GHz
-    - Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **Inference Mode Description**
-
-| Mode        | GPU Configuration                        | CPU Configuration | Acceleration Technology Combination                   |
-|-------------|----------------------------------------|-------------------|---------------------------------------------------|
-| Normal Mode | FP32 Precision / No TRT Acceleration   | FP32 Precision / 8 Threads | PaddleInference                                 |
-| High-Performance Mode | Optimal combination of pre-selected precision types and acceleration strategies | FP32 Precision / 8 Threads | Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.) |
+<strong>Test Environment Description:</strong>
+
+  <ul>
+      <li><b>Performance Test Environment</b>
+          <ul>
+          <li><strong>Test Dataset:</strong> PaddleX Internal Self-built Evaluation Dataset.</li>
+              <li><strong>Hardware Configuration:</strong>
+                  <ul>
+                      <li>GPU: NVIDIA Tesla T4</li>
+                      <li>CPU: Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>Inference Mode Description</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>Mode</th>
+            <th>GPU Configuration </th>
+            <th>CPU Configuration </th>
+            <th>Acceleration Technology Combination</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>Normal Mode</td>
+            <td>FP32 Precision / No TRT Acceleration</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>High-Performance Mode</td>
+            <td>Optimal combination of pre-selected precision types and acceleration strategies</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.)</td>
+        </tr>
+    </tbody>
+</table>
 
 
 ## III. Quick Integration
@@ -126,6 +153,20 @@ Relevant methods, parameters, and explanations are as follows:
 <td>None</td>
 <td>None</td>
 </tr>
+<tr>
+<td><code>device</code></td>
+<td>The device used for model inference</td>
+<td><code>str</code></td>
+<td>It supports specifying specific GPU card numbers, such as "gpu:0", other hardware card numbers, such as "npu:0", or CPU, such as "cpu".</td>
+<td><code>gpu:0</code></td>
+</tr>
+<tr>
+<td><code>use_hpip</code></td>
+<td>Whether to enable high-performance inference. </td>
+<td><code>bool</code></td>
+<td>None</td>
+<td><code>False</code></td>
+</tr>
 </table>
 
 * <code>model_name</code> must be specified. After specifying <code>model_name</code>, the default model parameters from PaddleX will be used. If <code>model_dir</code> is specified, the user-defined model will be used.

+ 49 - 15
docs/module_usage/tutorials/ocr_modules/table_structure_recognition.md

@@ -50,21 +50,48 @@ comments: true
 </tr>
 </table>
 
-**测试环境说明:**
-
-- **性能测试环境**
-  - **测试数据集**:PaddleX 内部自建高难度中文表格识别数据集。
-  - **硬件配置**:
-    - GPU:NVIDIA Tesla T4
-    - CPU:Intel Xeon Gold 6271C @ 2.60GHz
-    - 其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **推理模式说明**
-
-| 模式        | GPU配置                          | CPU配置          | 加速技术组合                                |
-|-------------|----------------------------------|------------------|---------------------------------------------|
-| 常规模式    | FP32精度 / 无TRT加速             | FP32精度 / 8线程       | PaddleInference                             |
-| 高性能模式  | 选择先验精度类型和加速策略的最优组合         | FP32精度 / 8线程       | 选择先验最优后端(Paddle/OpenVINO/TRT等) |
+<strong>测试环境说明:</strong>
+
+  <ul>
+      <li><b>性能测试环境</b>
+          <ul>
+              <li><strong>测试数据集:</strong>PaddleX 内部自建高难度中文表格识别数据集。</li>
+              <li><strong>硬件配置:</strong>
+                  <ul>
+                      <li>GPU:NVIDIA Tesla T4</li>
+                      <li>CPU:Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>推理模式说明</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>模式</th>
+            <th>GPU配置</th>
+            <th>CPU配置</th>
+            <th>加速技术组合</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>常规模式</td>
+            <td>FP32精度 / 无TRT加速</td>
+            <td>FP32精度 / 8线程</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>高性能模式</td>
+            <td>选择先验精度类型和加速策略的最优组合</td>
+            <td>FP32精度 / 8线程</td>
+            <td>选择先验最优后端(Paddle/OpenVINO/TRT等)</td>
+        </tr>
+    </tbody>
+</table>
 
 
 ## 三、快速集成
@@ -122,6 +149,13 @@ for res in output:
 <td>无</td>
 </tr>
 <tr>
+<td><code>device</code></td>
+<td>模型推理设备</td>
+<td><code>str</code></td>
+<td>支持指定GPU具体卡号,如“gpu:0”,其他硬件具体卡号,如“npu:0”,CPU如“cpu”。</td>
+<td><code>gpu:0</code></td>
+</tr>
+<tr>
 <td><code>use_hpip</code></td>
 <td>是否启用高性能推理</td>
 <td><code>bool</code></td>

+ 56 - 15
docs/module_usage/tutorials/ocr_modules/text_detection.en.md

@@ -39,21 +39,48 @@ The text detection module is a crucial component in OCR (Optical Character Recog
 </tbody>
 </table>
 
-**Test Environment Description**:
-
-- **Performance Test Environment**
-  - **Test Dataset**: PaddleOCR Self-built Dataset for Chinese and English, Covering Various Scenarios
-  - **Hardware Configuration**:
-    - GPU: NVIDIA Tesla T4
-    - CPU: Intel Xeon Gold 6271C @ 2.60GHz
-    - Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **Inference Mode Description**
-
-| Mode        | GPU Configuration                        | CPU Configuration | Acceleration Technology Combination                   |
-|-------------|----------------------------------------|-------------------|---------------------------------------------------|
-| Normal Mode | FP32 Precision / No TRT Acceleration   | FP32 Precision / 8 Threads | PaddleInference                                 |
-| High-Performance Mode | Optimal combination of pre-selected precision types and acceleration strategies | FP32 Precision / 8 Threads | Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.) |
+<strong>Test Environment Description:</strong>
+
+  <ul>
+      <li><b>Performance Test Environment</b>
+          <ul>
+           <li><strong>Test Dataset:</strong> PaddleOCR’s self-built Chinese and English dataset covers multiple scenarios including street views, web images, documents, and handwriting. The Chinese text detection dataset contains 593 images.</li>
+              <li><strong>Hardware Configuration:</strong>
+                  <ul>
+                      <li>GPU: NVIDIA Tesla T4</li>
+                      <li>CPU: Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>Inference Mode Description</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>Mode</th>
+            <th>GPU Configuration </th>
+            <th>CPU Configuration </th>
+            <th>Acceleration Technology Combination</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>Normal Mode</td>
+            <td>FP32 Precision / No TRT Acceleration</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>High-Performance Mode</td>
+            <td>Optimal combination of pre-selected precision types and acceleration strategies</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.)</td>
+        </tr>
+    </tbody>
+</table>
 
 ## III. Quick Integration
 > ❗ Before quick integration, please install the PaddleX wheel package. For detailed instructions, refer to the [PaddleX Local Installation Guide](../../../installation/installation.en.md).
@@ -116,6 +143,13 @@ Relevant methods, parameters, and explanations are as follows:
 <td>None</td>
 </tr>
 <tr>
+<td><code>device</code></td>
+<td>The device used for model inference</td>
+<td><code>str</code></td>
+<td>It supports specifying specific GPU card numbers, such as "gpu:0", other hardware card numbers, such as "npu:0", or CPU, such as "cpu".</td>
+<td><code>gpu:0</code></td>
+</tr>
+<tr>
 <td><code>limit_side_len</code></td>
 <td>Limit on the side length of the detection image</td>
 <td><code>int/None</code></td>
@@ -171,6 +205,13 @@ Relevant methods, parameters, and explanations are as follows:
 
 <td>None</td>
 </tr>
+<tr>
+<td><code>use_hpip</code></td>
+<td>Whether to enable high-performance inference. </td>
+<td><code>bool</code></td>
+<td>None</td>
+<td><code>False</code></td>
+</tr>
 </table>
 
 * The `model_name` must be specified. After specifying `model_name`, the default model parameters built into PaddleX will be used. If `model_dir` is specified, the user-defined model will be used.

+ 49 - 15
docs/module_usage/tutorials/ocr_modules/text_detection.md

@@ -57,21 +57,48 @@ comments: true
 </tbody>
 </table>
 
-**测试环境说明:**
-
-- **性能测试环境**
-  - **测试数据集**:PaddleOCR 自建的中英文数据集,覆盖街景、网图、文档、手写多个场景,其中文本识别包含 593 张图片。
-  - **硬件配置**:
-    - GPU:NVIDIA Tesla T4
-    - CPU:Intel Xeon Gold 6271C @ 2.60GHz
-    - 其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **推理模式说明**
-
-| 模式        | GPU配置                          | CPU配置          | 加速技术组合                                |
-|-------------|----------------------------------|------------------|---------------------------------------------|
-| 常规模式    | FP32精度 / 无TRT加速             | FP32精度 / 8线程       | PaddleInference                             |
-| 高性能模式  | 选择先验精度类型和加速策略的最优组合         | FP32精度 / 8线程       | 选择先验最优后端(Paddle/OpenVINO/TRT等) |
+<strong>测试环境说明:</strong>
+
+  <ul>
+      <li><b>性能测试环境</b>
+          <ul>
+              <li><strong>测试数据集:</strong>PaddleOCR 自建的中英文数据集,覆盖街景、网图、文档、手写多个场景,其中文本检测包含 593 张图片。</li>
+              <li><strong>硬件配置:</strong>
+                  <ul>
+                      <li>GPU:NVIDIA Tesla T4</li>
+                      <li>CPU:Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>推理模式说明</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>模式</th>
+            <th>GPU配置</th>
+            <th>CPU配置</th>
+            <th>加速技术组合</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>常规模式</td>
+            <td>FP32精度 / 无TRT加速</td>
+            <td>FP32精度 / 8线程</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>高性能模式</td>
+            <td>选择先验精度类型和加速策略的最优组合</td>
+            <td>FP32精度 / 8线程</td>
+            <td>选择先验最优后端(Paddle/OpenVINO/TRT等)</td>
+        </tr>
+    </tbody>
+</table>
 
 ## 三、快速集成
 > ❗ 在快速集成前,请先安装 PaddleX 的 wheel 包,详细请参考 [PaddleX本地安装教程](../../../installation/installation.md)。
@@ -132,6 +159,13 @@ for res in output:
 <td>无</td>
 </tr>
 <tr>
+<td><code>device</code></td>
+<td>模型推理设备</td>
+<td><code>str</code></td>
+<td>支持指定GPU具体卡号,如“gpu:0”,其他硬件具体卡号,如“npu:0”,CPU如“cpu”。</td>
+<td><code>gpu:0</code></td>
+</tr>
+<tr>
 <td><code>limit_side_len</code></td>
 <td>检测的图像边长限制</td>
 <td><code>int/None</code></td>

+ 55 - 13
docs/module_usage/tutorials/ocr_modules/text_image_unwarping.en.md

@@ -28,21 +28,49 @@ The primary purpose of Text Image Unwarping is to perform geometric transformati
 </tbody>
 </table>
 
-**Test Environment Description**:
+<strong>Test Environment Description:</strong>
 
-- **Performance Test Environment**
-  - **Test Dataset**: [DocUNet benchmark](https://www3.cs.stonybrook.edu/~cvl/docunet.html) dataset.
-  - **Hardware Configuration**:
-    - GPU: NVIDIA Tesla T4
-    - CPU: Intel Xeon Gold 6271C @ 2.60GHz
-    - Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
+  <ul>
+      <li><b>Performance Test Environment</b>
+          <ul>
+          <li><strong>Test Dataset:</strong>
+          <a href="https://www3.cs.stonybrook.edu/~cvl/docunet.html">DocUNet benchmark</a> dataset.</li>
+              <li><strong>Hardware Configuration:</strong>
+                  <ul>
+                      <li>GPU: NVIDIA Tesla T4</li>
+                      <li>CPU: Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>Inference Mode Description</b></li>
+  </ul>
 
-- **Inference Mode Description**
-
-| Mode        | GPU Configuration                        | CPU Configuration | Acceleration Technology Combination                   |
-|-------------|----------------------------------------|-------------------|---------------------------------------------------|
-| Normal Mode | FP32 Precision / No TRT Acceleration   | FP32 Precision / 8 Threads | PaddleInference                                 |
-| High-Performance Mode | Optimal combination of pre-selected precision types and acceleration strategies | FP32 Precision / 8 Threads | Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.) |
+<table border="1">
+    <thead>
+        <tr>
+            <th>Mode</th>
+            <th>GPU Configuration </th>
+            <th>CPU Configuration </th>
+            <th>Acceleration Technology Combination</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>Normal Mode</td>
+            <td>FP32 Precision / No TRT Acceleration</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>High-Performance Mode</td>
+            <td>Optimal combination of pre-selected precision types and acceleration strategies</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.)</td>
+        </tr>
+    </tbody>
+</table>
 
 ## III. Quick Integration
 > ❗ Before quick integration, please install the PaddleX wheel package. For detailed instructions, refer to the [PaddleX Local Installation Guide](../../../installation/installation.en.md)
@@ -105,6 +133,20 @@ Relevant methods, parameters, and explanations are as follows:
 <td>None</td>
 <td>None</td>
 </tr>
+<tr>
+<td><code>device</code></td>
+<td>The device used for model inference</td>
+<td><code>str</code></td>
+<td>It supports specifying specific GPU card numbers, such as "gpu:0", other hardware card numbers, such as "npu:0", or CPU, such as "cpu".</td>
+<td><code>gpu:0</code></td>
+</tr>
+<tr>
+<td><code>use_hpip</code></td>
+<td>Whether to enable high-performance inference. </td>
+<td><code>bool</code></td>
+<td>None</td>
+<td><code>False</code></td>
+</tr>
 </table>
 
 * The `model_name` must be specified. After specifying `model_name`, the default model parameters built into PaddleX will be used. If `model_dir` is specified, the user-defined model will be used.

+ 49 - 15
docs/module_usage/tutorials/ocr_modules/text_image_unwarping.md

@@ -30,21 +30,48 @@ comments: true
 </tbody>
 </table>
 
-**测试环境说明:**
-
-- **性能测试环境**
-  - **测试数据集**:[DocUNet benchmark](https://www3.cs.stonybrook.edu/~cvl/docunet.html)数据集。
-  - **硬件配置**:
-    - GPU:NVIDIA Tesla T4
-    - CPU:Intel Xeon Gold 6271C @ 2.60GHz
-    - 其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **推理模式说明**
-
-| 模式        | GPU配置                          | CPU配置          | 加速技术组合                                |
-|-------------|----------------------------------|------------------|---------------------------------------------|
-| 常规模式    | FP32精度 / 无TRT加速             | FP32精度 / 8线程       | PaddleInference                             |
-| 高性能模式  | 选择先验精度类型和加速策略的最优组合         | FP32精度 / 8线程       | 选择先验最优后端(Paddle/OpenVINO/TRT等) |
+<strong>测试环境说明:</strong>
+
+  <ul>
+      <li><b>性能测试环境</b>
+          <ul>
+              <li><strong>测试数据集:</strong><a href="https://www3.cs.stonybrook.edu/~cvl/docunet.html">DocUNet benchmark</a>数据集。</li>
+              <li><strong>硬件配置:</strong>
+                  <ul>
+                      <li>GPU:NVIDIA Tesla T4</li>
+                      <li>CPU:Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>推理模式说明</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>模式</th>
+            <th>GPU配置</th>
+            <th>CPU配置</th>
+            <th>加速技术组合</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>常规模式</td>
+            <td>FP32精度 / 无TRT加速</td>
+            <td>FP32精度 / 8线程</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>高性能模式</td>
+            <td>选择先验精度类型和加速策略的最优组合</td>
+            <td>FP32精度 / 8线程</td>
+            <td>选择先验最优后端(Paddle/OpenVINO/TRT等)</td>
+        </tr>
+    </tbody>
+</table>
 
 
 ## 三、快速集成
@@ -103,6 +130,13 @@ for res in output:
 <td>无</td>
 </tr>
 <tr>
+<td><code>device</code></td>
+<td>模型推理设备</td>
+<td><code>str</code></td>
+<td>支持指定GPU具体卡号,如“gpu:0”,其他硬件具体卡号,如“npu:0”,CPU如“cpu”。</td>
+<td><code>gpu:0</code></td>
+</tr>
+<tr>
 <td><code>use_hpip</code></td>
 <td>是否启用高性能推理</td>
 <td><code>bool</code></td>

+ 57 - 20
docs/module_usage/tutorials/ocr_modules/text_recognition.en.md

@@ -262,26 +262,56 @@ The ultra-lightweight cyrillic alphabet recognition model trained based on the P
 </tr>
 </table>
 
-**Test Environment Description**:
-
-- **Performance Test Environment**
-  - **Test Dataset**:
-    - Chinese Recognition Model: A self-built Chinese dataset using PaddleOCR, covering various scenarios such as street scenes, web images, documents, and handwriting, with 11,000 images for text recognition.
-    - ch_SVTRv2_rec: <a href="https://aistudio.baidu.com/competition/detail/1131/0/introduction">PaddleOCR Algorithm Model Challenge - Track 1: OCR End-to-End Recognition Task</a> A-Rank Evaluation Set.
-    - ch_RepSVTR_rec: <a href="https://aistudio.baidu.com/competition/detail/1131/0/introduction">PaddleOCR Algorithm Model Challenge - Track 1: OCR End-to-End Recognition Task</a> B-Rank Evaluation Set.
-    - English Recognition Model: A self-built English dataset using PaddleX.
-    - Multilingual Recognition Model: A self-built multilingual dataset using PaddleX.
-  - **Hardware Configuration**:
-    - GPU: NVIDIA Tesla T4
-    - CPU: Intel Xeon Gold 6271C @ 2.60GHz
-    - Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **Inference Mode Description**
-
-| Mode        | GPU Configuration                        | CPU Configuration | Acceleration Technology Combination                   |
-|-------------|----------------------------------------|-------------------|---------------------------------------------------|
-| Normal Mode | FP32 Precision / No TRT Acceleration   | FP32 Precision / 8 Threads | PaddleInference                                 |
-| High-Performance Mode | Optimal combination of pre-selected precision types and acceleration strategies | FP32 Precision / 8 Threads | Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.) |
+<strong>Test Environment Description:</strong>
+
+  <ul>
+      <li><b>Performance Test Environment</b>
+          <ul>
+             <li><strong>Test Dataset:</strong>
+               <ul>
+                 <li>Chinese Recognition Model: A self-built Chinese dataset using PaddleOCR, covering various scenarios such as street scenes, web images, documents, and handwriting, with 11,000 images for text recognition.</li>
+                 <li>ch_SVTRv2_rec: <a href="https://aistudio.baidu.com/competition/detail/1131/0/introduction">PaddleOCR Algorithm Model Challenge - Track 1: OCR End-to-End Recognition Task</a> A-Rank Evaluation Set.</li>
+                 <li>ch_RepSVTR_rec: <a href="https://aistudio.baidu.com/competition/detail/1131/0/introduction">PaddleOCR Algorithm Model Challenge - Track 1: OCR End-to-End Recognition Task</a> B-Rank Evaluation Set.</li>
+                 <li>English Recognition Model: A self-built English dataset using PaddleX.</li>
+                 <li>Multilingual Recognition Model: A self-built multilingual dataset using PaddleX.</li>
+               </ul>
+             </li>
+              <li><strong>Hardware Configuration:</strong>
+                  <ul>
+                      <li>GPU: NVIDIA Tesla T4</li>
+                      <li>CPU: Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>Inference Mode Description</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>Mode</th>
+            <th>GPU Configuration </th>
+            <th>CPU Configuration </th>
+            <th>Acceleration Technology Combination</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>Normal Mode</td>
+            <td>FP32 Precision / No TRT Acceleration</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>High-Performance Mode</td>
+            <td>Optimal combination of pre-selected precision types and acceleration strategies</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.)</td>
+        </tr>
+    </tbody>
+</table>
 
 </details>
 
@@ -344,6 +374,13 @@ The explanations for the methods, parameters, etc., are as follows:
 <td>None</td>
 </tr>
 <tr>
+<td><code>device</code></td>
+<td>The device used for model inference</td>
+<td><code>str</code></td>
+<td>It supports specifying specific GPU card numbers, such as "gpu:0", other hardware card numbers, such as "npu:0", or CPU, such as "cpu".</td>
+<td><code>gpu:0</code></td>
+</tr>
+<tr>
 <td><code>use_hpip</code></td>
 <td>Whether to enable high-performance inference. </td>
 <td><code>bool</code></td>

+ 67 - 20
docs/module_usage/tutorials/ocr_modules/text_recognition.md

@@ -277,26 +277,66 @@ devanagari_PP-OCRv3_mobile_rec_infer.tar">推理模型</a>/<a href="https://padd
 </tr>
 </table>
 
-**测试环境说明:**
-
-- **性能测试环境**
-  - **测试数据集**:
-    - 中文识别模型: PaddleOCR 自建的中文数据集,覆盖街景、网图、文档、手写多个场景,其中文本识别包含 1.1w 张图片。
-    - ch_SVTRv2_rec:<a href="https://aistudio.baidu.com/competition/detail/1131/0/introduction">PaddleOCR算法模型挑战赛 - 赛题一:OCR端到端识别任务</a>A榜评估集。
-    - ch_RepSVTR_rec:<a href="https://aistudio.baidu.com/competition/detail/1131/0/introduction">PaddleOCR算法模型挑战赛 - 赛题一:OCR端到端识别任务</a>B榜评估集。
-    - 英文识别模型:PaddleX 自建的英文数据集。
-    - 多语言识别模型:PaddleX 自建的多语种数据集。
-  - **硬件配置**:
-    - GPU:NVIDIA Tesla T4
-    - CPU:Intel Xeon Gold 6271C @ 2.60GHz
-    - 其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **推理模式说明**
-
-| 模式        | GPU配置                          | CPU配置          | 加速技术组合                                |
-|-------------|----------------------------------|------------------|---------------------------------------------|
-| 常规模式    | FP32精度 / 无TRT加速             | FP32精度 / 8线程       | PaddleInference                             |
-| 高性能模式  | 选择先验精度类型和加速策略的最优组合         | FP32精度 / 8线程       | 选择先验最优后端(Paddle/OpenVINO/TRT等) |
+<strong>测试环境说明:</strong>
+
+  <ul>
+      <li><b>性能测试环境</b>
+          <ul>
+              <li><strong>测试数据集:</strong>
+                 <ul>
+                    <li>
+                    中文识别模型: PaddleOCR 自建的中文数据集,覆盖街景、网图、文档、手写多个场景,其中文本识别包含 1.1w 张图片。
+                    </li>
+                    <li>
+                      ch_SVTRv2_rec:<a href="https://aistudio.baidu.com/competition/detail/1131/0/introduction">PaddleOCR算法模型挑战赛 - 赛题一:OCR端到端识别任务</a>A榜评估集。
+                    </li>
+                    <li>
+                      ch_RepSVTR_rec:<a href="https://aistudio.baidu.com/competition/detail/1131/0/introduction">PaddleOCR算法模型挑战赛 - 赛题一:OCR端到端识别任务</a>B榜评估集。
+                    </li>
+                    <li>
+                      英文识别模型:PaddleX 自建的英文数据集。
+                    </li>
+                    <li>
+                      多语言识别模型:PaddleX 自建的多语种数据集。
+                    </li>
+                 </ul>
+              </li>
+              <li><strong>硬件配置:</strong>
+                  <ul>
+                      <li>GPU:NVIDIA Tesla T4</li>
+                      <li>CPU:Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>推理模式说明</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>模式</th>
+            <th>GPU配置</th>
+            <th>CPU配置</th>
+            <th>加速技术组合</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>常规模式</td>
+            <td>FP32精度 / 无TRT加速</td>
+            <td>FP32精度 / 8线程</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>高性能模式</td>
+            <td>选择先验精度类型和加速策略的最优组合</td>
+            <td>FP32精度 / 8线程</td>
+            <td>选择先验最优后端(Paddle/OpenVINO/TRT等)</td>
+        </tr>
+    </tbody>
+</table>
 
 </details>
 
@@ -357,6 +397,13 @@ for res in output:
 <td>无</td>
 </tr>
 <tr>
+<td><code>device</code></td>
+<td>模型推理设备</td>
+<td><code>str</code></td>
+<td>支持指定GPU具体卡号,如“gpu:0”,其他硬件具体卡号,如“npu:0”,CPU如“cpu”。</td>
+<td><code>gpu:0</code></td>
+</tr>
+<tr>
 <td><code>use_hpip</code></td>
 <td>是否启用高性能推理</td>
 <td><code>bool</code></td>

+ 56 - 15
docs/module_usage/tutorials/ocr_modules/textline_orientation_classification.en.md

@@ -32,21 +32,48 @@ The text line orientation classification module primarily distinguishes the orie
 </tbody>
 </table>
 
-**Test Environment Description**:
-
-- **Performance Test Environment**
-  - **Test Dataset**: PaddleX Self-built Dataset, Covering Multiple Scenarios Such as Documents and Certificates, Containing 1000 Images.
-  - **Hardware Configuration**:
-    - GPU: NVIDIA Tesla T4
-    - CPU: Intel Xeon Gold 6271C @ 2.60GHz
-    - Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **Inference Mode Description**
-
-| Mode        | GPU Configuration                        | CPU Configuration | Acceleration Technology Combination                   |
-|-------------|----------------------------------------|-------------------|---------------------------------------------------|
-| Normal Mode | FP32 Precision / No TRT Acceleration   | FP32 Precision / 8 Threads | PaddleInference                                 |
-| High-Performance Mode | Optimal combination of pre-selected precision types and acceleration strategies | FP32 Precision / 8 Threads | Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.) |
+<strong>Test Environment Description:</strong>
+
+  <ul>
+      <li><b>Performance Test Environment</b>
+          <ul>
+             <li><strong>Test Dataset:</strong> PaddleX Self-built Dataset, Covering Multiple Scenarios Such as Documents and Certificates, Containing 1000 Images.</li>
+              <li><strong>Hardware Configuration:</strong>
+                  <ul>
+                      <li>GPU: NVIDIA Tesla T4</li>
+                      <li>CPU: Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>Inference Mode Description</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>Mode</th>
+            <th>GPU Configuration </th>
+            <th>CPU Configuration </th>
+            <th>Acceleration Technology Combination</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>Normal Mode</td>
+            <td>FP32 Precision / No TRT Acceleration</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>High-Performance Mode</td>
+            <td>Optimal combination of pre-selected precision types and acceleration strategies</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.)</td>
+        </tr>
+    </tbody>
+</table>
 
 ## III. Quick Integration
 
@@ -108,6 +135,20 @@ The explanations for the methods, parameters, etc., are as follows:
 <td>None</td>
 <td>None</td>
 </tr>
+<tr>
+<td><code>device</code></td>
+<td>The device used for model inference</td>
+<td><code>str</code></td>
+<td>It supports specifying specific GPU card numbers, such as "gpu:0", other hardware card numbers, such as "npu:0", or CPU, such as "cpu".</td>
+<td><code>gpu:0</code></td>
+</tr>
+<tr>
+<td><code>use_hpip</code></td>
+<td>Whether to enable high-performance inference. </td>
+<td><code>bool</code></td>
+<td>None</td>
+<td><code>False</code></td>
+</tr>
 </table>
 
 * The `model_name` must be specified. After specifying `model_name`, the default model parameters built into PaddleX are used. If `model_dir` is specified, the user-defined model is used.

+ 57 - 15
docs/module_usage/tutorials/ocr_modules/textline_orientation_classification.md

@@ -34,21 +34,49 @@ comments: true
 </tbody>
 </table>
 
-**测试环境说明:**
-
-- **性能测试环境**
-  - **测试数据集**:PaddleX 自建的数据集,覆盖证件和文档等多个场景,包含 1000 张图片。
-  - **硬件配置**:
-    - GPU:NVIDIA Tesla T4
-    - CPU:Intel Xeon Gold 6271C @ 2.60GHz
-    - 其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **推理模式说明**
-
-| 模式        | GPU配置                          | CPU配置          | 加速技术组合                                |
-|-------------|----------------------------------|------------------|---------------------------------------------|
-| 常规模式    | FP32精度 / 无TRT加速             | FP32精度 / 8线程       | PaddleInference                             |
-| 高性能模式  | 选择先验精度类型和加速策略的最优组合         | FP32精度 / 8线程       | 选择先验最优后端(Paddle/OpenVINO/TRT等) |
+<strong>测试环境说明:</strong>
+
+  <ul>
+      <li><b>性能测试环境</b>
+          <ul>
+              <li><strong>测试数据集:</strong>PaddleX 自建的数据集,覆盖证件和文档等多个场景,包含 1000 张图片。</li>
+              <li><strong>硬件配置:</strong>
+                  <ul>
+                      <li>GPU:NVIDIA Tesla T4</li>
+                      <li>CPU:Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>推理模式说明</b></li>
+  </ul>
+
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>模式</th>
+            <th>GPU配置</th>
+            <th>CPU配置</th>
+            <th>加速技术组合</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>常规模式</td>
+            <td>FP32精度 / 无TRT加速</td>
+            <td>FP32精度 / 8线程</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>高性能模式</td>
+            <td>选择先验精度类型和加速策略的最优组合</td>
+            <td>FP32精度 / 8线程</td>
+            <td>选择先验最优后端(Paddle/OpenVINO/TRT等)</td>
+        </tr>
+    </tbody>
+</table>
 
 ## 三、快速集成
 
@@ -109,6 +137,20 @@ for res in output:
 <td>无</td>
 <td>无</td>
 </tr>
+<tr>
+<td><code>device</code></td>
+<td>模型推理设备</td>
+<td><code>str</code></td>
+<td>支持指定GPU具体卡号,如“gpu:0”,其他硬件具体卡号,如“npu:0”,CPU如“cpu”。</td>
+<td><code>gpu:0</code></td>
+</tr>
+<tr>
+<td><code>use_hpip</code></td>
+<td>是否启用高性能推理</td>
+<td><code>bool</code></td>
+<td>无</td>
+<td><code>False</code></td>
+</tr>
 </table>
 
 * 其中,`model_name` 必须指定,指定 `model_name` 后,默认使用 PaddleX 内置的模型参数,在此基础上,指定 `model_dir` 时,使用用户自定义的模型。

+ 14 - 0
docs/module_usage/tutorials/speech_modules/multilingual_speech_recognition.en.md

@@ -66,6 +66,20 @@ Related methods, parameters, and explanations are as follows:
 <td>None</td>
 <td>None</td>
 </tr>
+<tr>
+<td><code>device</code></td>
+<td>The device used for model inference</td>
+<td><code>str</code></td>
+<td>It supports specifying specific GPU card numbers, such as "gpu:0", other hardware card numbers, such as "npu:0", or CPU, such as "cpu".</td>
+<td><code>gpu:0</code></td>
+</tr>
+<tr>
+<td><code>use_hpip</code></td>
+<td>Whether to enable high-performance inference. </td>
+<td><code>bool</code></td>
+<td>None</td>
+<td><code>False</code></td>
+</tr>
 </table>
 
 * The `model_name` must be specified. After specifying `model_name`, the built-in model parameters of PaddleX are used by default. If `model_dir` is specified, the user-defined model is used.

+ 14 - 0
docs/module_usage/tutorials/speech_modules/multilingual_speech_recognition.md

@@ -115,6 +115,20 @@ for res in output:
 <td>无</td>
 <td>无</td>
 </tr>
+<tr>
+<td><code>device</code></td>
+<td>模型推理设备</td>
+<td><code>str</code></td>
+<td>支持指定GPU具体卡号,如“gpu:0”,其他硬件具体卡号,如“npu:0”,CPU如“cpu”。</td>
+<td><code>gpu:0</code></td>
+</tr>
+<tr>
+<td><code>use_hpip</code></td>
+<td>是否启用高性能推理</td>
+<td><code>bool</code></td>
+<td>无</td>
+<td><code>False</code></td>
+</tr>
 </table>
 
 * 其中,`model_name` 必须指定,指定 `model_name` 后,默认使用 PaddleX 内置的模型参数,在此基础上,指定 `model_dir` 时,使用用户自定义的模型。

+ 57 - 16
docs/module_usage/tutorials/time_series_modules/time_series_anomaly_detection.en.md

@@ -55,24 +55,51 @@ Time series anomaly detection focuses on identifying abnormal points or periods
 </tr>
 </tbody>
 </table>
-
-**Test Environment Description**:
-
-- **Performance Test Environment**
-  - **Test Dataset**: </b>PSM<b>dataset.
-  - **Hardware Configuration**:
-    - GPU: NVIDIA Tesla T4
-    - CPU: Intel Xeon Gold 6271C @ 2.60GHz
-    - Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **Inference Mode Description**
-
-| Mode        | GPU Configuration                        | CPU Configuration | Acceleration Technology Combination                   |
-|-------------|----------------------------------------|-------------------|---------------------------------------------------|
-| Normal Mode | FP32 Precision / No TRT Acceleration   | FP32 Precision / 8 Threads | PaddleInference                                 |
-| High-Performance Mode | Optimal combination of pre-selected precision types and acceleration strategies | FP32 Precision / 8 Threads | Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.) |
+<strong>Test Environment Description:</strong>
+
+  <ul>
+      <li><b>Performance Test Environment</b>
+          <ul>
+            <li><strong>Test Dataset:</strong> <b>PSM</b> dataset.</li>
+              <li><strong>Hardware Configuration:</strong>
+                  <ul>
+                      <li>GPU: NVIDIA Tesla T4</li>
+                      <li>CPU: Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>Inference Mode Description</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>Mode</th>
+            <th>GPU Configuration </th>
+            <th>CPU Configuration </th>
+            <th>Acceleration Technology Combination</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>Normal Mode</td>
+            <td>FP32 Precision / No TRT Acceleration</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>High-Performance Mode</td>
+            <td>Optimal combination of pre-selected precision types and acceleration strategies</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.)</td>
+        </tr>
+    </tbody>
+</table>
 
 ## III. Quick Integration
+
 > ❗ Before quick integration, please install the PaddleX wheel package. For details, refer to the [PaddleX Local Installation Guide](../../../installation/installation.en.md)
 
 After installing the wheel package, a few lines of code can complete the inference of the time series anomaly detection module. You can switch models under this module freely, and you can also integrate the model inference of the time series anomaly detection module into your project. Before running the following code, please download the [demo csv](https://paddle-model-ecology.bj.bcebos.com/paddlex/ts/demo_ts/ts_ad.csv) to your local machine.
@@ -138,6 +165,20 @@ Relevant methods, parameters, and explanations are as follows:
 <td>None</td>
 <td>None</td>
 </tr>
+<tr>
+<td><code>device</code></td>
+<td>The device used for model inference</td>
+<td><code>str</code></td>
+<td>It supports specifying specific GPU card numbers, such as "gpu:0", other hardware card numbers, such as "npu:0", or CPU, such as "cpu".</td>
+<td><code>gpu:0</code></td>
+</tr>
+<tr>
+<td><code>use_hpip</code></td>
+<td>Whether to enable high-performance inference. </td>
+<td><code>bool</code></td>
+<td>None</td>
+<td><code>False</code></td>
+</tr>
 </table>
 
 * Note that `model_name` must be specified. After specifying `model_name`, the default PaddleX built-in model parameters will be used. If `model_dir` is specified, the user-defined model will be used.

+ 49 - 15
docs/module_usage/tutorials/time_series_modules/time_series_anomaly_detection.md

@@ -58,21 +58,48 @@ comments: true
 </table>
 
 
-**测试环境说明:**
-
-- **性能测试环境**
-  - **测试数据集**:</b>PSM<b>数据集,时序输入长度为100。
-  - **硬件配置**:
-    - GPU:NVIDIA Tesla T4
-    - CPU:Intel Xeon Gold 6271C @ 2.60GHz
-    - 其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **推理模式说明**
-
-| 模式        | GPU配置                          | CPU配置          | 加速技术组合                                |
-|-------------|----------------------------------|------------------|---------------------------------------------|
-| 常规模式    | FP32精度 / 无TRT加速             | FP32精度 / 8线程       | PaddleInference                             |
-| 高性能模式  | 选择先验精度类型和加速策略的最优组合         | FP32精度 / 8线程       | 选择先验最优后端(Paddle/OpenVINO/TRT等) |
+<strong>测试环境说明:</strong>
+
+  <ul>
+      <li><b>性能测试环境</b>
+          <ul>
+          <li><strong>测试数据集:</strong><b>PSM</b>数据集,时序输入长度为100。</li>
+              <li><strong>硬件配置:</strong>
+                  <ul>
+                      <li>GPU:NVIDIA Tesla T4</li>
+                      <li>CPU:Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>推理模式说明</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>模式</th>
+            <th>GPU配置</th>
+            <th>CPU配置</th>
+            <th>加速技术组合</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>常规模式</td>
+            <td>FP32精度 / 无TRT加速</td>
+            <td>FP32精度 / 8线程</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>高性能模式</td>
+            <td>选择先验精度类型和加速策略的最优组合</td>
+            <td>FP32精度 / 8线程</td>
+            <td>选择先验最优后端(Paddle/OpenVINO/TRT等)</td>
+        </tr>
+    </tbody>
+</table>
 
 
 ## 三、快速集成
@@ -143,6 +170,13 @@ timestamp
 <td>无</td>
 </tr>
 <tr>
+<td><code>device</code></td>
+<td>模型推理设备</td>
+<td><code>str</code></td>
+<td>支持指定GPU具体卡号,如“gpu:0”,其他硬件具体卡号,如“npu:0”,CPU如“cpu”。</td>
+<td><code>gpu:0</code></td>
+</tr>
+<tr>
 <td><code>use_hpip</code></td>
 <td>是否启用高性能推理</td>
 <td><code>bool</code></td>

+ 56 - 15
docs/module_usage/tutorials/time_series_modules/time_series_classification.en.md

@@ -28,21 +28,48 @@ Time series classification involves identifying and categorizing different patte
 </tbody>
 </table>
 
-**Test Environment Description**:
-
-- **Performance Test Environment**
-  - **Test Dataset**: UWaveGestureLibrary.
-  - **Hardware Configuration**:
-    - GPU: NVIDIA Tesla T4
-    - CPU: Intel Xeon Gold 6271C @ 2.60GHz
-    - Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **Inference Mode Description**
-
-| Mode        | GPU Configuration                        | CPU Configuration | Acceleration Technology Combination                   |
-|-------------|----------------------------------------|-------------------|---------------------------------------------------|
-| Normal Mode | FP32 Precision / No TRT Acceleration   | FP32 Precision / 8 Threads | PaddleInference                                 |
-| High-Performance Mode | Optimal combination of pre-selected precision types and acceleration strategies | FP32 Precision / 8 Threads | Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.) |
+<strong>Test Environment Description:</strong>
+
+  <ul>
+      <li><b>Performance Test Environment</b>
+          <ul>
+           <li><strong>Test Dataset:</strong> UWaveGestureLibrary.</li>
+              <li><strong>Hardware Configuration:</strong>
+                  <ul>
+                      <li>GPU: NVIDIA Tesla T4</li>
+                      <li>CPU: Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>Inference Mode Description</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>Mode</th>
+            <th>GPU Configuration </th>
+            <th>CPU Configuration </th>
+            <th>Acceleration Technology Combination</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>Normal Mode</td>
+            <td>FP32 Precision / No TRT Acceleration</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>High-Performance Mode</td>
+            <td>Optimal combination of pre-selected precision types and acceleration strategies</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.)</td>
+        </tr>
+    </tbody>
+</table>
 
 ## III. Quick Integration
 > ❗ Before quick integration, please install the PaddleX wheel package. For detailed instructions, refer to [PaddleX Local Installation Guide](../../../installation/installation.en.md)
@@ -106,6 +133,20 @@ Descriptions of related methods, parameters, etc., are as follows:
 <td>None</td>
 <td>None</td>
 </tr>
+<tr>
+<td><code>device</code></td>
+<td>The device used for model inference</td>
+<td><code>str</code></td>
+<td>It supports specifying specific GPU card numbers, such as "gpu:0", other hardware card numbers, such as "npu:0", or CPU, such as "cpu".</td>
+<td><code>gpu:0</code></td>
+</tr>
+<tr>
+<td><code>use_hpip</code></td>
+<td>Whether to enable high-performance inference. </td>
+<td><code>bool</code></td>
+<td>None</td>
+<td><code>False</code></td>
+</tr>
 </table>
 
 * The `model_name` must be specified. After specifying `model_name`, PaddleX's built-in model parameters are used by default. If `model_dir` is specified, the user-defined model is used.

+ 49 - 15
docs/module_usage/tutorials/time_series_modules/time_series_classification.md

@@ -29,21 +29,48 @@ comments: true
 </tbody>
 </table>
 
-**测试环境说明:**
-
-- **性能测试环境**
-  - **测试数据集**:UWaveGestureLibrary评估集。
-  - **硬件配置**:
-    - GPU:NVIDIA Tesla T4
-    - CPU:Intel Xeon Gold 6271C @ 2.60GHz
-    - 其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **推理模式说明**
-
-| 模式        | GPU配置                          | CPU配置          | 加速技术组合                                |
-|-------------|----------------------------------|------------------|---------------------------------------------|
-| 常规模式    | FP32精度 / 无TRT加速             | FP32精度 / 8线程       | PaddleInference                             |
-| 高性能模式  | 选择先验精度类型和加速策略的最优组合         | FP32精度 / 8线程       | 选择先验最优后端(Paddle/OpenVINO/TRT等) |
+<strong>测试环境说明:</strong>
+
+  <ul>
+      <li><b>性能测试环境</b>
+          <ul>
+           <li><strong>测试数据集:</strong>UWaveGestureLibrary评估集。</li>
+              <li><strong>硬件配置:</strong>
+                  <ul>
+                      <li>GPU:NVIDIA Tesla T4</li>
+                      <li>CPU:Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>推理模式说明</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>模式</th>
+            <th>GPU配置</th>
+            <th>CPU配置</th>
+            <th>加速技术组合</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>常规模式</td>
+            <td>FP32精度 / 无TRT加速</td>
+            <td>FP32精度 / 8线程</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>高性能模式</td>
+            <td>选择先验精度类型和加速策略的最优组合</td>
+            <td>FP32精度 / 8线程</td>
+            <td>选择先验最优后端(Paddle/OpenVINO/TRT等)</td>
+        </tr>
+    </tbody>
+</table>
 
 
 ## 三、快速集成
@@ -110,6 +137,13 @@ for res in output:
 <td>无</td>
 </tr>
 <tr>
+<td><code>device</code></td>
+<td>模型推理设备</td>
+<td><code>str</code></td>
+<td>支持指定GPU具体卡号,如“gpu:0”,其他硬件具体卡号,如“npu:0”,CPU如“cpu”。</td>
+<td><code>gpu:0</code></td>
+</tr>
+<tr>
 <td><code>use_hpip</code></td>
 <td>是否启用高性能推理</td>
 <td><code>bool</code></td>

+ 56 - 15
docs/module_usage/tutorials/time_series_modules/time_series_forecasting.en.md

@@ -58,21 +58,48 @@ Time series forecasting aims to predict the possible values or states at a futur
 </tbody>
 </table>
 
-**Test Environment Description**:
-
-- **Performance Test Environment**
-  - **Test Dataset**: [ETTH1](https://paddle-model-ecology.bj.bcebos.com/paddlex/data/Etth1.tar) test dataset.
-  - **Hardware Configuration**:
-    - GPU: NVIDIA Tesla T4
-    - CPU: Intel Xeon Gold 6271C @ 2.60GHz
-    - Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **Inference Mode Description**
-
-| Mode        | GPU Configuration                        | CPU Configuration | Acceleration Technology Combination                   |
-|-------------|----------------------------------------|-------------------|---------------------------------------------------|
-| Normal Mode | FP32 Precision / No TRT Acceleration   | FP32 Precision / 8 Threads | PaddleInference                                 |
-| High-Performance Mode | Optimal combination of pre-selected precision types and acceleration strategies | FP32 Precision / 8 Threads | Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.) |
+<strong>Test Environment Description:</strong>
+
+  <ul>
+      <li><b>Performance Test Environment</b>
+          <ul>
+            <li><strong>Test Dataset:</strong> <a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/data/Etth1.tar">ETTH1</a> test dataset.</li>
+              <li><strong>Hardware Configuration:</strong>
+                  <ul>
+                      <li>GPU: NVIDIA Tesla T4</li>
+                      <li>CPU: Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>Inference Mode Description</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>Mode</th>
+            <th>GPU Configuration </th>
+            <th>CPU Configuration </th>
+            <th>Acceleration Technology Combination</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>Normal Mode</td>
+            <td>FP32 Precision / No TRT Acceleration</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>High-Performance Mode</td>
+            <td>Optimal combination of pre-selected precision types and acceleration strategies</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.)</td>
+        </tr>
+    </tbody>
+</table>
 
 ## III. Quick Integration
 > ❗ Before quick integration, please install the PaddleX wheel package. For detailed instructions, refer to the [PaddleX Local Installation Guide](../../../installation/installation.en.md)
@@ -141,6 +168,20 @@ Descriptions of related methods, parameters, etc., are as follows:
 <td>None</td>
 <td>None</td>
 </tr>
+<tr>
+<td><code>device</code></td>
+<td>The device used for model inference</td>
+<td><code>str</code></td>
+<td>It supports specifying specific GPU card numbers, such as "gpu:0", other hardware card numbers, such as "npu:0", or CPU, such as "cpu".</td>
+<td><code>gpu:0</code></td>
+</tr>
+<tr>
+<td><code>use_hpip</code></td>
+<td>Whether to enable high-performance inference. </td>
+<td><code>bool</code></td>
+<td>None</td>
+<td><code>False</code></td>
+</tr>
 </table>
 
 * The `model_name` must be specified. After specifying `model_name`, PaddleX's built-in model parameters are used by default. If `model_dir` is specified, the user-defined model is used.

+ 49 - 15
docs/module_usage/tutorials/time_series_modules/time_series_forecasting.md

@@ -72,21 +72,48 @@ comments: true
 </tbody>
 </table>
 
-**测试环境说明:**
-
-- **性能测试环境**
-  - **测试数据集**:[ETTH1](https://paddle-model-ecology.bj.bcebos.com/paddlex/data/Etth1.tar)<b>测试数据集。
-  - **硬件配置**:
-    - GPU:NVIDIA Tesla T4
-    - CPU:Intel Xeon Gold 6271C @ 2.60GHz
-    - 其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **推理模式说明**
-
-| 模式        | GPU配置                          | CPU配置          | 加速技术组合                                |
-|-------------|----------------------------------|------------------|---------------------------------------------|
-| 常规模式    | FP32精度 / 无TRT加速             | FP32精度 / 8线程       | PaddleInference                             |
-| 高性能模式  | 选择先验精度类型和加速策略的最优组合         | FP32精度 / 8线程       | 选择先验最优后端(Paddle/OpenVINO/TRT等) |
+<strong>测试环境说明:</strong>
+
+  <ul>
+      <li><b>性能测试环境</b>
+          <ul>
+           <li><strong>测试数据集:</strong><a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/data/Etth1.tar">ETTH1</a>测试数据集。</li>
+              <li><strong>硬件配置:</strong>
+                  <ul>
+                      <li>GPU:NVIDIA Tesla T4</li>
+                      <li>CPU:Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>推理模式说明</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>模式</th>
+            <th>GPU配置</th>
+            <th>CPU配置</th>
+            <th>加速技术组合</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>常规模式</td>
+            <td>FP32精度 / 无TRT加速</td>
+            <td>FP32精度 / 8线程</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>高性能模式</td>
+            <td>选择先验精度类型和加速策略的最优组合</td>
+            <td>FP32精度 / 8线程</td>
+            <td>选择先验最优后端(Paddle/OpenVINO/TRT等)</td>
+        </tr>
+    </tbody>
+</table>
 
 
 ## 三、快速集成
@@ -157,6 +184,13 @@ date
 <td>无</td>
 </tr>
 <tr>
+<td><code>device</code></td>
+<td>模型推理设备</td>
+<td><code>str</code></td>
+<td>支持指定GPU具体卡号,如“gpu:0”,其他硬件具体卡号,如“npu:0”,CPU如“cpu”。</td>
+<td><code>gpu:0</code></td>
+</tr>
+<tr>
 <td><code>use_hpip</code></td>
 <td>是否启用高性能推理</td>
 <td><code>bool</code></td>

+ 56 - 15
docs/module_usage/tutorials/video_modules/video_classification.en.md

@@ -40,21 +40,48 @@ PP-TSM is a video classification model developed by Baidu PaddlePaddle's Vision
 
 </table>
 
-**Test Environment Description**:
-
-- **Performance Test Environment**
-  - **Test Dataset**: <a href="https://github.com/PaddlePaddle/PaddleVideo/blob/develop/docs/zh-CN/dataset/k400.md">K400</a> validation set.
-  - **Hardware Configuration**:
-    - GPU: NVIDIA Tesla T4
-    - CPU: Intel Xeon Gold 6271C @ 2.60GHz
-    - Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **Inference Mode Description**
-
-| Mode        | GPU Configuration                        | CPU Configuration | Acceleration Technology Combination                   |
-|-------------|----------------------------------------|-------------------|---------------------------------------------------|
-| Normal Mode | FP32 Precision / No TRT Acceleration   | FP32 Precision / 8 Threads | PaddleInference                                 |
-| High-Performance Mode | Optimal combination of pre-selected precision types and acceleration strategies | FP32 Precision / 8 Threads | Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.) |
+<strong>Test Environment Description:</strong>
+
+  <ul>
+      <li><b>Performance Test Environment</b>
+          <ul>
+               <li><strong>Test Dataset:</strong> <a href="https://github.com/PaddlePaddle/PaddleVideo/blob/develop/docs/zh-CN/dataset/k400.md">K400</a> validation set.</li>
+              <li><strong>Hardware Configuration:</strong>
+                  <ul>
+                      <li>GPU: NVIDIA Tesla T4</li>
+                      <li>CPU: Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>Inference Mode Description</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>Mode</th>
+            <th>GPU Configuration </th>
+            <th>CPU Configuration </th>
+            <th>Acceleration Technology Combination</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>Normal Mode</td>
+            <td>FP32 Precision / No TRT Acceleration</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>High-Performance Mode</td>
+            <td>Optimal combination of pre-selected precision types and acceleration strategies</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.)</td>
+        </tr>
+    </tbody>
+</table>
 </details>
 
 ## III. Quick Integration
@@ -118,6 +145,20 @@ The Python script above performs the following steps:
 <td>None</td>
 </tr>
 <tr>
+<td><code>device</code></td>
+<td>The device used for model inference</td>
+<td><code>str</code></td>
+<td>It supports specifying specific GPU card numbers, such as "gpu:0", other hardware card numbers, such as "npu:0", or CPU, such as "cpu".</td>
+<td><code>gpu:0</code></td>
+</tr>
+<tr>
+<td><code>use_hpip</code></td>
+<td>Whether to enable high-performance inference. </td>
+<td><code>bool</code></td>
+<td>None</td>
+<td><code>False</code></td>
+</tr>
+<tr>
 <td><code> topk</code></td>
 <td>The top <code> topk</code> categories and corresponding classification probabilities of the prediction result;if not specified, the default configuration of the PaddleX official model will be used</td>
 <td><code>int</code></td>

+ 56 - 15
docs/module_usage/tutorials/video_modules/video_classification.md

@@ -40,21 +40,48 @@ PP-TSM是一种百度飞桨视觉团队自研的视频分类模型。该模型
 
 </table>
 
-**测试环境说明:**
-
-- **性能测试环境**
-  - **测试数据集**:<a href="https://github.com/PaddlePaddle/PaddleVideo/blob/develop/docs/zh-CN/dataset/k400.md">K400</a> 验证集。
-  - **硬件配置**:
-    - GPU:NVIDIA Tesla T4
-    - CPU:Intel Xeon Gold 6271C @ 2.60GHz
-    - 其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **推理模式说明**
-
-| 模式        | GPU配置                          | CPU配置          | 加速技术组合                                |
-|-------------|----------------------------------|------------------|---------------------------------------------|
-| 常规模式    | FP32精度 / 无TRT加速             | FP32精度 / 8线程       | PaddleInference                             |
-| 高性能模式  | 选择先验精度类型和加速策略的最优组合         | FP32精度 / 8线程       | 选择先验最优后端(Paddle/OpenVINO/TRT等) |
+<strong>测试环境说明:</strong>
+
+  <ul>
+      <li><b>性能测试环境</b>
+          <ul>
+            <li><strong>测试数据集:</strong><a href="https://github.com/PaddlePaddle/PaddleVideo/blob/develop/docs/zh-CN/dataset/k400.md">K400</a> 验证集。</li>
+              <li><strong>硬件配置:</strong>
+                  <ul>
+                      <li>GPU:NVIDIA Tesla T4</li>
+                      <li>CPU:Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>推理模式说明</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>模式</th>
+            <th>GPU配置</th>
+            <th>CPU配置</th>
+            <th>加速技术组合</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>常规模式</td>
+            <td>FP32精度 / 无TRT加速</td>
+            <td>FP32精度 / 8线程</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>高性能模式</td>
+            <td>选择先验精度类型和加速策略的最优组合</td>
+            <td>FP32精度 / 8线程</td>
+            <td>选择先验最优后端(Paddle/OpenVINO/TRT等)</td>
+        </tr>
+    </tbody>
+</table>
 
 </details>
 
@@ -116,6 +143,20 @@ for res in output:
 <td>无</td>
 </tr>
 <tr>
+<td><code>device</code></td>
+<td>模型推理设备</td>
+<td><code>str</code></td>
+<td>支持指定GPU具体卡号,如“gpu:0”,其他硬件具体卡号,如“npu:0”,CPU如“cpu”。</td>
+<td><code>gpu:0</code></td>
+</tr>
+<tr>
+<td><code>use_hpip</code></td>
+<td>是否启用高性能推理</td>
+<td><code>bool</code></td>
+<td>无</td>
+<td><code>False</code></td>
+</tr>
+<tr>
 <td><code> topk</code></td>
 <td>预测结果的前 <code>topk</code> 个类别和对应的分类概率;如果不指定,将默认使用PaddleX官方模型配置</td>
 <td><code>int</code></td>

+ 56 - 15
docs/module_usage/tutorials/video_modules/video_detection.en.md

@@ -30,21 +30,48 @@ YOWO is a single-stage network with two branches. One branch extracts spatial fe
 
 </table>
 
-**Test Environment Description**:
-
-- **Performance Test Environment**
-  - **Test Dataset**: <a href="http://www.thumos.info/download.html">UCF101-24</a> test set.
-  - **Hardware Configuration**:
-    - GPU: NVIDIA Tesla T4
-    - CPU: Intel Xeon Gold 6271C @ 2.60GHz
-    - Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **Inference Mode Description**
-
-| Mode        | GPU Configuration                        | CPU Configuration | Acceleration Technology Combination                   |
-|-------------|----------------------------------------|-------------------|---------------------------------------------------|
-| Normal Mode | FP32 Precision / No TRT Acceleration   | FP32 Precision / 8 Threads | PaddleInference                                 |
-| High-Performance Mode | Optimal combination of pre-selected precision types and acceleration strategies | FP32 Precision / 8 Threads | Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.) |
+<strong>Test Environment Description:</strong>
+
+  <ul>
+      <li><b>Performance Test Environment</b>
+          <ul>
+               <li><strong>Test Dataset:</strong> <a href="http://www.thumos.info/download.html">UCF101-24</a> test set.</li>
+              <li><strong>Hardware Configuration:</strong>
+                  <ul>
+                      <li>GPU: NVIDIA Tesla T4</li>
+                      <li>CPU: Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>Inference Mode Description</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>Mode</th>
+            <th>GPU Configuration </th>
+            <th>CPU Configuration </th>
+            <th>Acceleration Technology Combination</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>Normal Mode</td>
+            <td>FP32 Precision / No TRT Acceleration</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>High-Performance Mode</td>
+            <td>Optimal combination of pre-selected precision types and acceleration strategies</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.)</td>
+        </tr>
+    </tbody>
+</table>
 
 ## <span id="lable">III. Quick Integration</span>
 > ❗ Before quick integration, please install the PaddleX wheel package. For detailed instructions, refer to the [PaddleX Local Installation Guide](../../../installation/installation.en.md).
@@ -108,6 +135,20 @@ In the above Python script, the following steps are executed:
 <td>None</td>
 </tr>
 <tr>
+<td><code>device</code></td>
+<td>The device used for model inference</td>
+<td><code>str</code></td>
+<td>It supports specifying specific GPU card numbers, such as "gpu:0", other hardware card numbers, such as "npu:0", or CPU, such as "cpu".</td>
+<td><code>gpu:0</code></td>
+</tr>
+<tr>
+<td><code>use_hpip</code></td>
+<td>Whether to enable high-performance inference. </td>
+<td><code>bool</code></td>
+<td>None</td>
+<td><code>False</code></td>
+</tr>
+<tr>
 <td><code>nms_thresh</code></td>
 <td>The IoU threshold parameter in the Non-Maximum Suppression (NMS) process; if not specified, the default configuration of the PaddleX official model will be used</td>
 <td><code>float/None</code></td>

+ 56 - 15
docs/module_usage/tutorials/video_modules/video_detection.md

@@ -29,21 +29,48 @@ YOWO是具有两个分支的单阶段网络。一个分支通过2D-CNN提取关
 
 </table>
 
-**测试环境说明:**
-
-- **性能测试环境**
-  - **测试数据集**:<a href="http://www.thumos.info/download.html">UCF101-24</a> test数据集。
-  - **硬件配置**:
-    - GPU:NVIDIA Tesla T4
-    - CPU:Intel Xeon Gold 6271C @ 2.60GHz
-    - 其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **推理模式说明**
-
-| 模式        | GPU配置                          | CPU配置          | 加速技术组合                                |
-|-------------|----------------------------------|------------------|---------------------------------------------|
-| 常规模式    | FP32精度 / 无TRT加速             | FP32精度 / 8线程       | PaddleInference                             |
-| 高性能模式  | 选择先验精度类型和加速策略的最优组合         | FP32精度 / 8线程       | 选择先验最优后端(Paddle/OpenVINO/TRT等) |
+<strong>测试环境说明:</strong>
+
+  <ul>
+      <li><b>性能测试环境</b>
+          <ul>
+              <li><strong>测试数据集:</strong><a href="http://www.thumos.info/download.html">UCF101-24</a> test数据集。</li>
+              <li><strong>硬件配置:</strong>
+                  <ul>
+                      <li>GPU:NVIDIA Tesla T4</li>
+                      <li>CPU:Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>推理模式说明</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>模式</th>
+            <th>GPU配置</th>
+            <th>CPU配置</th>
+            <th>加速技术组合</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>常规模式</td>
+            <td>FP32精度 / 无TRT加速</td>
+            <td>FP32精度 / 8线程</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>高性能模式</td>
+            <td>选择先验精度类型和加速策略的最优组合</td>
+            <td>FP32精度 / 8线程</td>
+            <td>选择先验最优后端(Paddle/OpenVINO/TRT等)</td>
+        </tr>
+    </tbody>
+</table>
 
 </details>
 
@@ -110,6 +137,20 @@ for res in output:
 <td>无</td>
 </tr>
 <tr>
+<td><code>device</code></td>
+<td>模型推理设备</td>
+<td><code>str</code></td>
+<td>支持指定GPU具体卡号,如“gpu:0”,其他硬件具体卡号,如“npu:0”,CPU如“cpu”。</td>
+<td><code>gpu:0</code></td>
+</tr>
+<tr>
+<td><code>use_hpip</code></td>
+<td>是否启用高性能推理</td>
+<td><code>bool</code></td>
+<td>无</td>
+<td><code>False</code></td>
+</tr>
+<tr>
 <td><code> nms_thresh</code></td>
 <td>非极大值抑制(Non-Maximum Suppression, NMS)过程中的IoU阈值参数;如果不指定,将默认使用PaddleX官方模型配置</td>
 <td><code>float/None</code></td>

+ 44 - 17
docs/pipeline_usage/tutorials/cv_pipelines/3d_bev_detection.en.md

@@ -32,21 +32,48 @@ BEVFusion is a multi-modal 3D object detection model that fuses surround camera
 <tr>
 </table>
 
-**Test Environment Description**:
-
-- **Performance Test Environment**
-  - **Test Dataset**: <a href="https://www.nuscenes.org/nuscenes">nuscenes</a> validation set
-  - **Hardware Configuration**:
-    - GPU: NVIDIA Tesla T4
-    - CPU: Intel Xeon Gold 6271C @ 2.60GHz
-    - Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **Inference Mode Description**
-
-| Mode        | GPU Configuration                        | CPU Configuration | Acceleration Technology Combination                   |
-|-------------|----------------------------------------|-------------------|---------------------------------------------------|
-| Normal Mode | FP32 Precision / No TRT Acceleration   | FP32 Precision / 8 Threads | PaddleInference                                 |
-| High-Performance Mode | Optimal combination of pre-selected precision types and acceleration strategies | FP32 Precision / 8 Threads | Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.) |
+<strong>Test Environment Description:</strong>
+
+  <ul>
+      <li><b>Performance Test Environment</b>
+          <ul>
+              <li><strong>Test Dataset:</strong><a href="https://www.nuscenes.org/nuscenes">nuscenes</a> validation set</li>
+              <li><strong>Hardware Configuration:</strong>
+                  <ul>
+                      <li>GPU: NVIDIA Tesla T4</li>
+                      <li>CPU: Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>Inference Mode Description</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>Mode</th>
+            <th>GPU Configuration </th>
+            <th>CPU Configuration </th>
+            <th>Acceleration Technology Combination</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>Normal Mode</td>
+            <td>FP32 Precision / No TRT Acceleration</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>High-Performance Mode</td>
+            <td>Optimal combination of pre-selected precision types and acceleration strategies</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.)</td>
+        </tr>
+    </tbody>
+</table>
 
 </details>
 
@@ -168,7 +195,7 @@ pip install open3d
 python paddlex/inference/models/3d_bev_detection/visualizer_3d.py --save_path="./output/"
 ```
 
-<img src="https://raw.githubusercontent.com/cuicheng01/PaddleX_doc_images/refs/heads/main/images/images/pipelines/3d_bev_detection/02.png">
+<img src="https://raw.githubusercontent.com/cuicheng01/PaddleX_doc_images/main/images/pipelines/3d_bev_detection/02.png">
 
 In the above Python script, the following steps are executed:
 
@@ -470,7 +497,7 @@ If the default model weights provided by the 3D multi-modal fusion detection pip
 
 ### 4.1 Model Fine-Tuning
 
-Refer to the [Custom Development](../../../module_usage/tutorials/cv_modules/3d_bev_detection.md#四二次开发) section in the [3D Multi-modal Fusion Detection Module Development Tutorial](../../../module_usage/tutorials/cv_modules/3d_bev_detection.md) and use your private dataset to fine-tune the model.
+Refer to the <b>Custom Development</b> section in the [3D Multi-modal Fusion Detection Module Development Tutorial](https://paddlepaddle.github.io/PaddleX/latest/en/module_usage/tutorials/cv_modules/3d_bev_detection.html) and use your private dataset to fine-tune the model.
 
 ### 4.2 Model Application
 After completing fine-tuning training using your private dataset, you will obtain local model weight files.

+ 44 - 17
docs/pipeline_usage/tutorials/cv_pipelines/3d_bev_detection.md

@@ -32,21 +32,48 @@ BEVFusion 是一种多模态 3D 目标检测模型,通过将环视摄像头图
 <tr>
 </table>
 
-**测试环境说明:**
-
-- **性能测试环境**
-  - **测试数据集**:<a href="https://www.nuscenes.org/nuscenes">nuscenes</a>验证集
-  - **硬件配置**:
-    - GPU:NVIDIA Tesla T4
-    - CPU:Intel Xeon Gold 6271C @ 2.60GHz
-    - 其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **推理模式说明**
-
-| 模式        | GPU配置                          | CPU配置          | 加速技术组合                                |
-|-------------|----------------------------------|------------------|---------------------------------------------|
-| 常规模式    | FP32精度 / 无TRT加速             | FP32精度 / 8线程       | PaddleInference                             |
-| 高性能模式  | 选择先验精度类型和加速策略的最优组合         | FP32精度 / 8线程       | 选择先验最优后端(Paddle/OpenVINO/TRT等) |
+<strong>测试环境说明:</strong>
+
+  <ul>
+      <li><b>性能测试环境</b>
+          <ul>
+              <li><strong>测试数据集:</strong><a href="https://www.nuscenes.org/nuscenes">nuscenes</a>验证集</li>
+              <li><strong>硬件配置:</strong>
+                  <ul>
+                      <li>GPU:NVIDIA Tesla T4</li>
+                      <li>CPU:Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>推理模式说明</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>模式</th>
+            <th>GPU配置</th>
+            <th>CPU配置</th>
+            <th>加速技术组合</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>常规模式</td>
+            <td>FP32精度 / 无TRT加速</td>
+            <td>FP32精度 / 8线程</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>高性能模式</td>
+            <td>选择先验精度类型和加速策略的最优组合</td>
+            <td>FP32精度 / 8线程</td>
+            <td>选择先验最优后端(Paddle/OpenVINO/TRT等)</td>
+        </tr>
+    </tbody>
+</table>
 
 ## 2. 快速开始
 
@@ -158,7 +185,7 @@ pip install open3d
 python paddlex/inference/models/3d_bev_detection/visualizer_3d.py --save_path="./output/"
 ```
 
-<img src="https://raw.githubusercontent.com/cuicheng01/PaddleX_doc_images/refs/heads/main/images/images/pipelines/3d_bev_detection/02.png">
+<img src="https://raw.githubusercontent.com/cuicheng01/PaddleX_doc_images/main/images/pipelines/3d_bev_detection/02.png">
 
 
 在上述 Python 脚本中,执行了如下几个步骤:
@@ -462,7 +489,7 @@ print(result["detectedObjects"])
 
 ### 4.1 模型微调
 
-参考[3D多模态融合检测模块开发教程](../../../module_usage/tutorials/cv_modules/3d_bev_detection.md)中的[二次开发](../../../module_usage/tutorials/cv_modules/3d_bev_detection.md#四二次开发)章节,使用您的私有数据集模型进行微调。
+参考[3D多模态融合检测模块开发教程](https://paddlepaddle.github.io/PaddleX/latest/module_usage/tutorials/cv_modules/3d_bev_detection.html)中的<b>二次开发</b>章节,使用您的私有数据集模型进行微调。
 
 ### 4.2 模型应用
 当您使用私有数据集完成微调训练后,可获得本地模型权重文件。

+ 48 - 18
docs/pipeline_usage/tutorials/cv_pipelines/face_recognition.en.md

@@ -96,23 +96,53 @@ The face recognition pipeline is an end-to-end system dedicated to solving face
 </tbody>
 </table>
 
-**Test Environment Description**:
-
-- **Performance Test Environment**
-  - **Test Dataset**:
-    - Face Detection Model: Evaluated on the WIDER-FACE validation set in COCO format with an input size of 640*640.
-    - Face Feature Model: Evaluated on the AgeDB-30, CFP-FP, and LFW datasets, respectively.
-  - **Hardware Configuration**:
-    - GPU: NVIDIA Tesla T4
-    - CPU: Intel Xeon Gold 6271C @ 2.60GHz
-    - Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **Inference Mode Description**
-
-| Mode        | GPU Configuration                        | CPU Configuration | Acceleration Technology Combination                   |
-|-------------|----------------------------------------|-------------------|---------------------------------------------------|
-| Normal Mode | FP32 Precision / No TRT Acceleration   | FP32 Precision / 8 Threads | PaddleInference                                 |
-| High-Performance Mode | Optimal combination of pre-selected precision types and acceleration strategies | FP32 Precision / 8 Threads | Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.) |
+<strong>Test Environment Description:</strong>
+
+  <ul>
+      <li><b>Performance Test Environment</b>
+          <ul>
+            <li><strong>Test Dataset:</strong>
+              <ul>
+                <li>Face Detection Model: Evaluated on the WIDER-FACE validation set in COCO format with an input size of 640*640.</li>
+                <li>Face Feature Model: Evaluated on the AgeDB-30, CFP-FP, and LFW datasets, respectively.</li>
+              </ul>
+            </li>
+              <li><strong>Hardware Configuration:</strong>
+                  <ul>
+                      <li>GPU: NVIDIA Tesla T4</li>
+                      <li>CPU: Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>Inference Mode Description</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>Mode</th>
+            <th>GPU Configuration </th>
+            <th>CPU Configuration </th>
+            <th>Acceleration Technology Combination</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>Normal Mode</td>
+            <td>FP32 Precision / No TRT Acceleration</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>High-Performance Mode</td>
+            <td>Optimal combination of pre-selected precision types and acceleration strategies</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.)</td>
+        </tr>
+    </tbody>
+</table>
 
 ## 2. Quick Start
 The pre-trained model pipelines provided by PaddleX can be quickly experienced. You can experience the effects of the face recognition pipeline online or locally using command-line or Python.
@@ -1072,7 +1102,7 @@ If the default model weights provided by the face recognition pipeline do not me
 ### 4.1 Model Fine-Tuning
 Since the face recognition pipeline includes two modules (face detection and face feature), the unsatisfactory performance of the model pipeline may come from either module.
 
-You can analyze the images with poor recognition performance. If you find that many faces are not detected during the analysis, it may indicate a deficiency in the face detection model. You need to refer to the [Face Detection Module Development Tutorial](../../../module_usage/tutorials/cv_modules/face_detection.en.md) and the [Custom Development](../../../module_usage/tutorials/cv_modules/face_detection.en.md) section to fine-tune the face detection model using your private dataset. If there are matching errors in the detected faces, it indicates that the face feature module needs further improvement. You need to refer to the [Face Feature Module Development Tutorial](../../../module_usage/tutorials/cv_modules/face_feature.md) and the [Custom Development](../../../module_usage/tutorials/cv_modules/face_feature.md) section to fine-tune the face feature module.
+You can analyze the images with poor recognition performance. If you find that many faces are not detected during the analysis, it may indicate a deficiency in the face detection model. You need to refer to the [Face Detection Module Development Tutorial](https://paddlepaddle.github.io/PaddleX/latest/en/module_usage/tutorials/cv_modules/face_detection.html) and the <b>Custom Development</b>section to fine-tune the face detection model using your private dataset. If there are matching errors in the detected faces, it indicates that the face feature module needs further improvement. You need to refer to the [Face Feature Module Development Tutorial](https://paddlepaddle.github.io/PaddleX/latest/en/module_usage/tutorials/cv_modules/face_feature.html) and the <b>Custom Development</b> section to fine-tune the face feature module.
 
 ### 4.2 Model Application
 After completing the fine-tuning training with your private dataset, you will obtain the local model weight file.

+ 49 - 18
docs/pipeline_usage/tutorials/cv_pipelines/face_recognition.md

@@ -96,23 +96,54 @@ comments: true
 </tbody>
 </table>
 
-**测试环境说明:**
-
-- **性能测试环境**
-  - **测试数据集**
-    - 人脸检测模型:COCO 格式的 WIDER-FACE 验证集上,以640*640作为输入尺寸评估得到的。
-    - 人脸特征模型:分别在 AgeDB-30、CFP-FP 和 LFW 数据集。
-  - **硬件配置**:
-    - GPU:NVIDIA Tesla T4
-    - CPU:Intel Xeon Gold 6271C @ 2.60GHz
-    - 其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **推理模式说明**
-
-| 模式        | GPU配置                          | CPU配置          | 加速技术组合                                |
-|-------------|----------------------------------|------------------|---------------------------------------------|
-| 常规模式    | FP32精度 / 无TRT加速             | FP32精度 / 8线程       | PaddleInference                             |
-| 高性能模式  | 选择先验精度类型和加速策略的最优组合         | FP32精度 / 8线程       | 选择先验最优后端(Paddle/OpenVINO/TRT等) |
+<strong>测试环境说明:</strong>
+
+  <ul>
+      <li><b>性能测试环境</b>
+          <ul>
+             <li><strong>测试数据集:
+             </strong>
+               <ul>
+                 <li>人脸检测模型:COCO 格式的 WIDER-FACE 验证集上,以640*640作为输入尺寸评估得到的。</li>
+                 <li>人脸特征模型:分别在 AgeDB-30、CFP-FP 和 LFW 数据集。</li>
+               </ul>
+             </li>
+              <li><strong>硬件配置:</strong>
+                  <ul>
+                      <li>GPU:NVIDIA Tesla T4</li>
+                      <li>CPU:Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>推理模式说明</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>模式</th>
+            <th>GPU配置</th>
+            <th>CPU配置</th>
+            <th>加速技术组合</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>常规模式</td>
+            <td>FP32精度 / 无TRT加速</td>
+            <td>FP32精度 / 8线程</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>高性能模式</td>
+            <td>选择先验精度类型和加速策略的最优组合</td>
+            <td>FP32精度 / 8线程</td>
+            <td>选择先验最优后端(Paddle/OpenVINO/TRT等)</td>
+        </tr>
+    </tbody>
+</table>
 
 ## 2. 快速开始
 PaddleX 所提供的模型产线均可以快速体验效果,你可以在线体验人脸识别产线的效果,也可以在本地使用命令行或 Python 体验人脸识别产线的效果。
@@ -1069,7 +1100,7 @@ pprint.pp(result_infer["faces"])
 ### 4.1 模型微调
 由于人脸识别产线包含两个模块(人脸检测和人脸特征),模型产线的效果不及预期可能来自于其中任何一个模块。
 
-您可以对识别效果差的图片进行分析,如果在分析过程中发现有较多的人脸未被检测出来,那么可能是人脸检测模型存在不足,您需要参考[人脸检测模块开发教程](../../../module_usage/tutorials/cv_modules/face_detection.md)中的[二次开发](../../../module_usage/tutorials/cv_modules/face_detection.md#四二次开发)章节,使用您的私有数据集对人脸检测模型进行微调;如果在已检测到的人脸出现匹配错误,这表明人脸特征模块需要进一步改进,您需要参考[人脸特征模块开发教程](../../../module_usage/tutorials/cv_modules/face_feature.md)中的[二次开发](../../../module_usage/tutorials/cv_modules/face_feature.md#四二次开发)章节,对人脸特征模块进行微调。
+您可以对识别效果差的图片进行分析,如果在分析过程中发现有较多的人脸未被检测出来,那么可能是人脸检测模型存在不足,您需要参考[人脸检测模块开发教程](https://paddlepaddle.github.io/PaddleX/latest/module_usage/tutorials/cv_modules/face_detection.html)中的<b>二次开发</b>章节,使用您的私有数据集对人脸检测模型进行微调;如果在已检测到的人脸出现匹配错误,这表明人脸特征模块需要进一步改进,您需要参考[人脸特征模块开发教程](https://paddlepaddle.github.io/PaddleX/latest/module_usage/tutorials/cv_modules/face_feature.html)中的<b>二次开发</b>章节,对人脸特征模块进行微调。
 
 ### 4.2 模型应用
 当您使用私有数据集完成微调训练后,可获得本地模型权重文件。

+ 48 - 18
docs/pipeline_usage/tutorials/cv_pipelines/general_image_recognition.en.md

@@ -71,23 +71,53 @@ PP-ShiTuV2 is a practical general image recognition system mainly composed of th
 </tr>
 </table>
 
-**Test Environment Description**:
-
-- **Performance Test Environment**
-  - **Test Dataset**:
-    - Subject Detection Model: PaddleClas Subject Detection Dataset.
-    - Image Feature Model: AliProducts Dataset.
-  - **Hardware Configuration**:
-    - GPU: NVIDIA Tesla T4
-    - CPU: Intel Xeon Gold 6271C @ 2.60GHz
-    - Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **Inference Mode Description**
-
-| Mode        | GPU Configuration                        | CPU Configuration | Acceleration Technology Combination                   |
-|-------------|----------------------------------------|-------------------|---------------------------------------------------|
-| Normal Mode | FP32 Precision / No TRT Acceleration   | FP32 Precision / 8 Threads | PaddleInference                                 |
-| High-Performance Mode | Optimal combination of pre-selected precision types and acceleration strategies | FP32 Precision / 8 Threads | Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.) |
+<strong>Test Environment Description:</strong>
+
+  <ul>
+      <li><b>Performance Test Environment</b>
+          <ul>
+            <li><strong>Test Dataset:</strong>
+              <ul>
+                <li>Subject Detection Model: PaddleClas Subject Detection Dataset.</li>
+                <li>Image Feature Model: AliProducts Dataset.</li>
+              </ul>
+            </li>
+              <li><strong>Hardware Configuration:</strong>
+                  <ul>
+                      <li>GPU: NVIDIA Tesla T4</li>
+                      <li>CPU: Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>Inference Mode Description</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>Mode</th>
+            <th>GPU Configuration </th>
+            <th>CPU Configuration </th>
+            <th>Acceleration Technology Combination</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>Normal Mode</td>
+            <td>FP32 Precision / No TRT Acceleration</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>High-Performance Mode</td>
+            <td>Optimal combination of pre-selected precision types and acceleration strategies</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.)</td>
+        </tr>
+    </tbody>
+</table>
 
 ## 2. Quick Start
 
@@ -1033,7 +1063,7 @@ If the default model weights provided by the general image recognition pipeline
 
 Since the general image recognition pipeline includes two modules (main body detection module and image feature module), the suboptimal performance of the model pipeline may come from either module.
 
-You can analyze the images with poor recognition results. If you find that many main body targets are not detected during the analysis, it may be due to the inadequacy of the main body detection model. You need to refer to the [Main Body Detection Module Development Tutorial](../../../module_usage/tutorials/cv_modules/mainbody_detection.en.md) in the [Custom Development](../../../module_usage/tutorials/cv_modules/mainbody_detection.en.md) section to fine-tune the main body detection model using your private dataset. If there are matching errors in the detected main bodies, it indicates that the image feature model needs further improvement. You need to refer to the [Image Feature Module Development Tutorial](../../../module_usage/tutorials/cv_modules/image_feature.en.md) in the [Custom Development](../../../module_usage/tutorials/cv_modules/image_feature.en.md) section to fine-tune the image feature model.
+You can analyze the images with poor recognition results. If you find that many main body targets are not detected during the analysis, it may be due to the inadequacy of the main body detection model. You need to refer to the [Main Body Detection Module Development Tutorial](https://paddlepaddle.github.io/PaddleX/latest/en/module_usage/tutorials/cv_modules/mainbody_detection.html) in the <b>Custom Development</b> section to fine-tune the main body detection model using your private dataset. If there are matching errors in the detected main bodies, it indicates that the image feature model needs further improvement. You need to refer to the [Image Feature Module Development Tutorial](https://paddlepaddle.github.io/PaddleX/latest/en/module_usage/tutorials/cv_modules/image_feature.html) in the <b>Custom Development</b> section to fine-tune the image feature model.
 
 ### 4.2 Model Application
 

+ 49 - 18
docs/pipeline_usage/tutorials/cv_pipelines/general_image_recognition.md

@@ -70,23 +70,54 @@ PP-ShiTuV2 是一个实用的通用图像识别系统,主要由主体检测、
 </tr>
 </table>
 
-**测试环境说明:**
-
-- **性能测试环境**
-  - **测试数据集**:
-    - 主体检测模型:PaddleClas 主体检测数据集。
-    - 图像特征模型:AliProducts数据集。
-  - **硬件配置**:
-    - GPU:NVIDIA Tesla T4
-    - CPU:Intel Xeon Gold 6271C @ 2.60GHz
-    - 其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **推理模式说明**
-
-| 模式        | GPU配置                          | CPU配置          | 加速技术组合                                |
-|-------------|----------------------------------|------------------|---------------------------------------------|
-| 常规模式    | FP32精度 / 无TRT加速             | FP32精度 / 8线程       | PaddleInference                             |
-| 高性能模式  | 选择先验精度类型和加速策略的最优组合         | FP32精度 / 8线程       | 选择先验最优后端(Paddle/OpenVINO/TRT等) |
+<strong>测试环境说明:</strong>
+
+  <ul>
+      <li><b>性能测试环境</b>
+          <ul>
+            <li><strong>测试数据集:
+             </strong>
+               <ul>
+                 <li>主体检测模型:PaddleClas 主体检测数据集。</li>
+                 <li>图像特征模型:AliProducts数据集。</li>
+               </ul>
+             </li>
+              <li><strong>硬件配置:</strong>
+                  <ul>
+                      <li>GPU:NVIDIA Tesla T4</li>
+                      <li>CPU:Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>推理模式说明</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>模式</th>
+            <th>GPU配置</th>
+            <th>CPU配置</th>
+            <th>加速技术组合</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>常规模式</td>
+            <td>FP32精度 / 无TRT加速</td>
+            <td>FP32精度 / 8线程</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>高性能模式</td>
+            <td>选择先验精度类型和加速策略的最优组合</td>
+            <td>FP32精度 / 8线程</td>
+            <td>选择先验最优后端(Paddle/OpenVINO/TRT等)</td>
+        </tr>
+    </tbody>
+</table>
 
 ## 2. 快速开始
 
@@ -1032,7 +1063,7 @@ pprint.pp(result_infer["detectedObjects"])
 
 由于通用图像识别产线包含两个模块(主体检测模块和图像特征模块),模型产线的效果不及预期可能来自于其中任何一个模块。
 
-您可以对识别效果差的图片进行分析,如果在分析过程中发现有较多的主体目标未被检测出来,那么可能是主体检测模型存在不足,您需要参考[主体检测模块开发教程](../../../module_usage/tutorials/cv_modules/mainbody_detection.md)中的[二次开发](../../../module_usage/tutorials/cv_modules/mainbody_detection.md#四二次开发)章节,使用您的私有数据集对主体检测模型进行微调;如果在已检测到的主体出现匹配错误,这表明图像特征模型需要进一步改进,您需要参考[图像特征模块开发教程](../../../module_usage/tutorials/cv_modules/image_feature.md)中的[二次开发](../../../module_usage/tutorials/cv_modules/image_feature.md#四二次开发)章节,对图像特征模型进行微调。
+您可以对识别效果差的图片进行分析,如果在分析过程中发现有较多的主体目标未被检测出来,那么可能是主体检测模型存在不足,您需要参考[主体检测模块开发教程](https://paddlepaddle.github.io/PaddleX/latest/module_usage/tutorials/cv_modules/mainbody_detection.html)中的<b>二次开发</b>章节,使用您的私有数据集对主体检测模型进行微调;如果在已检测到的主体出现匹配错误,这表明图像特征模型需要进一步改进,您需要参考[图像特征模块开发教程](https://paddlepaddle.github.io/PaddleX/latest/module_usage/tutorials/cv_modules/image_feature.html)中的<b>二次开发</b>章节,对图像特征模型进行微调。
 
 ### 4.2 模型应用
 

+ 48 - 18
docs/pipeline_usage/tutorials/cv_pipelines/human_keypoint_detection.en.md

@@ -76,23 +76,53 @@ PaddleX's Human Keypoint Detection Pipeline is a Top-Down solution consisting of
 </tr>
 </table>
 
-**Test Environment Description**:
-
-- **Performance Test Environment**
-  - **Test Dataset**:
-    - Pedestrian Detection Model: CrowdHuman Dataset.
-    - Human Keypoint Detection Model: COCO Dataset AP(0.5:0.95), with detection boxes obtained from ground truth annotations.
-  - **Hardware Configuration**:
-    - GPU: NVIDIA Tesla T4
-    - CPU: Intel Xeon Gold 6271C @ 2.60GHz
-    - Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **Inference Mode Description**
-
-| Mode        | GPU Configuration                        | CPU Configuration | Acceleration Technology Combination                   |
-|-------------|----------------------------------------|-------------------|---------------------------------------------------|
-| Normal Mode | FP32 Precision / No TRT Acceleration   | FP32 Precision / 8 Threads | PaddleInference                                 |
-| High-Performance Mode | Optimal combination of pre-selected precision types and acceleration strategies | FP32 Precision / 8 Threads | Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.) |
+<strong>Test Environment Description:</strong>
+
+  <ul>
+      <li><b>Performance Test Environment</b>
+          <ul>
+            <li><strong>Test Dataset:</strong>
+              <ul>
+                <li>Pedestrian Detection Model: CrowdHuman Dataset.</li>
+                <li>Human Keypoint Detection Model: COCO Dataset AP(0.5:0.95), with detection boxes obtained from ground truth annotations.</li>
+              </ul>
+            </li>
+              <li><strong>Hardware Configuration:</strong>
+                  <ul>
+                      <li>GPU: NVIDIA Tesla T4</li>
+                      <li>CPU: Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>Inference Mode Description</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>Mode</th>
+            <th>GPU Configuration </th>
+            <th>CPU Configuration </th>
+            <th>Acceleration Technology Combination</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>Normal Mode</td>
+            <td>FP32 Precision / No TRT Acceleration</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>High-Performance Mode</td>
+            <td>Optimal combination of pre-selected precision types and acceleration strategies</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.)</td>
+        </tr>
+    </tbody>
+</table>
 
 </details>
 
@@ -727,7 +757,7 @@ If the default model weights provided by the human keypoint detection pipeline d
 
 Since the human keypoint detection pipeline consists of two modules (pedestrian detection module and human keypoint detection module), the suboptimal performance of the model pipeline may stem from either module.
 
-You can analyze the images with poor recognition performance. If you find that many pedestrian targets are not detected during the analysis, it may indicate a deficiency in the pedestrian detection model. You need to refer to the [Pedestrian Detection Module Development Tutorial](../../../module_usage/tutorials/cv_modules/human_detection.en.md) in the [Custom Development](../../../module_usage/tutorials/cv_modules/human_detection.en.md) section to fine-tune the pedestrian detection model using your private dataset. If keypoint detection errors occur in detected pedestrians, it indicates that the keypoint detection model needs further improvement. You need to refer to the [Keypoint Detection Module Development Tutorial](../../../module_usage/tutorials/cv_modules/human_keypoint_detection.en.md) in the [Custom Development](../../../module_usage/tutorials/cv_modules/human_keypoint_detection.en.md#secondary-development) section to fine-tune the keypoint detection model.
+You can analyze the images with poor recognition performance. If you find that many pedestrian targets are not detected during the analysis, it may indicate a deficiency in the pedestrian detection model. You need to refer to the [Pedestrian Detection Module Development Tutorial](https://paddlepaddle.github.io/PaddleX/latest/en/module_usage/tutorials/cv_modules/human_detection.html) in the <b>Custom Development</b> section to fine-tune the pedestrian detection model using your private dataset. If keypoint detection errors occur in detected pedestrians, it indicates that the keypoint detection model needs further improvement. You need to refer to the [Keypoint Detection Module Development Tutorial](https://paddlepaddle.github.io/PaddleX/latest/en/module_usage/tutorials/cv_modules/human_keypoint_detection.html) in the <b>Custom Development</b> section to fine-tune the keypoint detection model.
 
 ### 4.2 Model Application
 

+ 24 - 12
docs/pipeline_usage/tutorials/cv_pipelines/human_keypoint_detection.md

@@ -73,18 +73,30 @@ PaddleX 的人体关键点检测产线是一个 Top-Down 方案,由行人检
 </tr>
 </table>
 
-<b>测试环境说明:</b>
+<strong>测试环境说明:</strong>
+
+  <ul>
+      <li><b>性能测试环境</b>
+          <ul>
+            <li><strong>测试数据集:
+             </strong>
+               <ul>
+                 <li>行人检测模型:CrowdHuman数据集。</li>
+                 <li>人体关键点检测模型:COCO数据集 AP(0.5:0.95),所依赖的检测框为ground truth标注得到。</li>
+               </ul>
+             </li>
+              <li><strong>硬件配置:</strong>
+                  <ul>
+                      <li>GPU:NVIDIA Tesla T4</li>
+                      <li>CPU:Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>推理模式说明</b></li>
+  </ul>
 
-- <b>性能测试环境</b>
-  - <b>测试数据集</b>:
-    - 行人检测模型:CrowdHuman数据集。
-    - 人体关键点检测模型:COCO数据集 AP(0.5:0.95),所依赖的检测框为ground truth标注得到。
-  - <b>硬件配置</b>:
-    - GPU:NVIDIA Tesla T4
-    - CPU:Intel Xeon Gold 6271C @ 2.60GHz
-    - 其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- <b>推理模式说明</b>
 
 <table border="1">
     <thead>
@@ -736,7 +748,7 @@ print(result["persons"])
 
 由于人体关键点检测产线包含两个模块(行人检测模块和人体关键点检测模块),模型产线的效果不及预期可能来自于其中任何一个模块。
 
-您可以对识别效果差的图片进行分析,如果在分析过程中发现有较多的行人目标未被检测出来,那么可能是行人检测模型存在不足,您需要参考[行人检测模块开发教程](../../../module_usage/tutorials/cv_modules/human_detection.md)中的[二次开发](../../../module_usage/tutorials/cv_modules/human_detection.md#四二次开发)章节,使用您的私有数据集对行人检测模型进行微调;如果在已检测到行人出现关键点检测错误,这表明关键点检测模型需要进一步改进,您需要参考[关键点检测模块开发教程](../../../module_usage/tutorials/cv_modules/human_keypoint_detection.md)中的[二次开发](../../../module_usage/tutorials/cv_modules/human_keypoint_detection.md#四二次开发)章节,对关键点检测模型进行微调。
+您可以对识别效果差的图片进行分析,如果在分析过程中发现有较多的行人目标未被检测出来,那么可能是行人检测模型存在不足,您需要参考[行人检测模块开发教程](https://paddlepaddle.github.io/PaddleX/latest/module_usage/tutorials/cv_modules/human_detection.html)中的<b>二次开发</b>章节,使用您的私有数据集对行人检测模型进行微调;如果在已检测到行人出现关键点检测错误,这表明关键点检测模型需要进一步改进,您需要参考[关键点检测模块开发教程](https://paddlepaddle.github.io/PaddleX/latest/module_usage/tutorials/cv_modules/human_keypoint_detection.html)中的<b>二次开发</b>章节,对关键点检测模型进行微调。
 
 ### 4.2 模型应用
 

+ 43 - 16
docs/pipeline_usage/tutorials/cv_pipelines/image_anomaly_detection.en.md

@@ -29,21 +29,48 @@ This pipeline integrates the high-precision anomaly detection model STFPM, which
 </tbody>
 </table>
 
-**Test Environment Description**:
-
-- **Performance Test Environment**
-  - **Test Dataset**: The above accuracy metrics are the average anomaly scores on the </b>[MVTec AD](https://www.mvtec.com/company/research/datasets/mvtec-ad)<b> validation set.
-  - **Hardware Configuration**:
-    - GPU: NVIDIA Tesla T4
-    - CPU: Intel Xeon Gold 6271C @ 2.60GHz
-    - Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **Inference Mode Description**
-
-| Mode        | GPU Configuration                        | CPU Configuration | Acceleration Technology Combination                   |
-|-------------|----------------------------------------|-------------------|---------------------------------------------------|
-| Normal Mode | FP32 Precision / No TRT Acceleration   | FP32 Precision / 8 Threads | PaddleInference                                 |
-| High-Performance Mode | Optimal combination of pre-selected precision types and acceleration strategies | FP32 Precision / 8 Threads | Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.) |
+<strong>Test Environment Description:</strong>
+
+  <ul>
+      <li><b>Performance Test Environment</b>
+          <ul>
+            <li><strong>Test Dataset:</strong>The above accuracy metrics are the average anomaly scores on the <b><a href="https://www.mvtec.com/company/research/datasets/mvtec-ad">MVTec AD</a></b> validation set.</li>
+              <li><strong>Hardware Configuration:</strong>
+                  <ul>
+                      <li>GPU: NVIDIA Tesla T4</li>
+                      <li>CPU: Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>Inference Mode Description</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>Mode</th>
+            <th>GPU Configuration </th>
+            <th>CPU Configuration </th>
+            <th>Acceleration Technology Combination</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>Normal Mode</td>
+            <td>FP32 Precision / No TRT Acceleration</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>High-Performance Mode</td>
+            <td>Optimal combination of pre-selected precision types and acceleration strategies</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.)</td>
+        </tr>
+    </tbody>
+</table>
 
 ## 2. Quick Start
 PaddleX provides pre-trained models for the anomaly detection pipeline, allowing for quick experience of its effects. You can use the command line or Python to experience the image anomaly detection pipeline locally.
@@ -798,7 +825,7 @@ You can choose the appropriate deployment method based on your needs to integrat
 If the default model weights provided by the image anomaly detection pipeline are not satisfactory in terms of accuracy or speed for your specific scenario, you can attempt to <b>further fine-tune the existing models using your own domain-specific or application-specific data</b> to improve the recognition performance of the image anomaly detection pipeline in your scenario.
 
 ### 4.1 Model Fine-Tuning
-Since the image anomaly detection pipeline includes an unsupervised image anomaly detection module, if the pipeline's performance does not meet expectations, you need to refer to the [Custom Development](../../../module_usage/tutorials/cv_modules/anomaly_detection.en.md) section in the [Unsupervised Anomaly Detection Module Development Guide](../../../module_usage/tutorials/cv_modules/anomaly_detection.en.md) and use your private dataset to fine-tune the image anomaly detection model.
+Since the image anomaly detection pipeline includes an unsupervised image anomaly detection module, if the pipeline's performance does not meet expectations, you need to refer to the <b>Custom Development</b> section in the [Unsupervised Anomaly Detection Module Development Guide](https://paddlepaddle.github.io/PaddleX/latest/en/module_usage/tutorials/cv_modules/anomaly_detection.html) and use your private dataset to fine-tune the image anomaly detection model.
 
 ### 4.2 Model Application
 After fine-tuning with your private dataset, you will obtain the local model weight file.

+ 46 - 16
docs/pipeline_usage/tutorials/cv_pipelines/image_anomaly_detection.md

@@ -30,21 +30,51 @@ comments: true
 </tbody>
 </table>
 
-**测试环境说明:**
-
-- **性能测试环境**
-  - **测试数据集**:<a href="https://www.mvtec.com/company/research/datasets/mvtec-ad">MVTec AD</a><b> 验证集 grid 数据。
-  - **硬件配置**:
-    - GPU:NVIDIA Tesla T4
-    - CPU:Intel Xeon Gold 6271C @ 2.60GHz
-    - 其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **推理模式说明**
-
-| 模式        | GPU配置                          | CPU配置          | 加速技术组合                                |
-|-------------|----------------------------------|------------------|---------------------------------------------|
-| 常规模式    | FP32精度 / 无TRT加速             | FP32精度 / 8线程       | PaddleInference                             |
-| 高性能模式  | 选择先验精度类型和加速策略的最优组合         | FP32精度 / 8线程       | 选择先验最优后端(Paddle/OpenVINO/TRT等) |
+<strong>测试环境说明:</strong>
+
+  <ul>
+      <li><b>性能测试环境</b>
+          <ul>
+          <li><strong>测试数据集:
+             </strong>
+              <b> <a href="https://www.mvtec.com/company/research/datasets/mvtec-ad">MVTec AD</a></b> 验证集 grid 数据。
+             </li>
+              <li><strong>硬件配置:</strong>
+                  <ul>
+                      <li>GPU:NVIDIA Tesla T4</li>
+                      <li>CPU:Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>推理模式说明</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>模式</th>
+            <th>GPU配置</th>
+            <th>CPU配置</th>
+            <th>加速技术组合</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>常规模式</td>
+            <td>FP32精度 / 无TRT加速</td>
+            <td>FP32精度 / 8线程</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>高性能模式</td>
+            <td>选择先验精度类型和加速策略的最优组合</td>
+            <td>FP32精度 / 8线程</td>
+            <td>选择先验最优后端(Paddle/OpenVINO/TRT等)</td>
+        </tr>
+    </tbody>
+</table>
 
 ## 2. 快速开始
 PaddleX 所提供的模型产线均可以快速体验效果,您可以在本地使用命令行或 Python 体验图像异常检测产线的效果。
@@ -798,7 +828,7 @@ echo &quot;Output image saved at &quot; . $output_image_path . &quot;\n&quot;;
 如果图像异常检测产线提供的默认模型权重在您的场景中,精度或速度不满意,您可以尝试利用<b>您自己拥有的特定领域或应用场景的数据</b>对现有模型进行进一步的<b>微调</b>,以提升图像异常检测产线的在您的场景中的识别效果。
 
 ### 4.1 模型微调
-由于图像异常检测产线包含无监督图像异常检测模块,如果模型产线的效果不及预期,那么您需要参考[无监督异常检测模块开发教程](../../../module_usage/tutorials/cv_modules/anomaly_detection.md)中的[二次开发](../../../module_usage/tutorials/cv_modules/anomaly_detection.md#四二次开发)章节,使用您的私有数据集对图像异常检测模型进行微调。
+由于图像异常检测产线包含无监督图像异常检测模块,如果模型产线的效果不及预期,那么您需要参考[无监督异常检测模块开发教程](https://paddlepaddle.github.io/PaddleX/latest/module_usage/tutorials/cv_modules/anomaly_detection.html)中的<b>二次开发</b>章节,使用您的私有数据集对图像异常检测模型进行微调。
 
 ### 4.2 模型应用
 当您使用私有数据集完成微调训练后,可获得本地模型权重文件。

+ 43 - 16
docs/pipeline_usage/tutorials/cv_pipelines/image_classification.en.md

@@ -685,21 +685,48 @@ Image classification is a technique that assigns images to predefined categories
 </tr>
 </tr></tr></tr></tr></table>
 
-**Test Environment Description**:
-
-- **Performance Test Environment**
-  - **Test Dataset**:  <a href="https://www.image-net.org/index.php">ImageNet-1k</a> validation set.
-  - **Hardware Configuration**:
-    - GPU: NVIDIA Tesla T4
-    - CPU: Intel Xeon Gold 6271C @ 2.60GHz
-    - Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **Inference Mode Description**
-
-| Mode        | GPU Configuration                        | CPU Configuration | Acceleration Technology Combination                   |
-|-------------|----------------------------------------|-------------------|---------------------------------------------------|
-| Normal Mode | FP32 Precision / No TRT Acceleration   | FP32 Precision / 8 Threads | PaddleInference                                 |
-| High-Performance Mode | Optimal combination of pre-selected precision types and acceleration strategies | FP32 Precision / 8 Threads | Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.) |
+<strong>Test Environment Description:</strong>
+
+  <ul>
+      <li><b>Performance Test Environment</b>
+          <ul>
+             <li><strong>Test Dataset:</strong><a href="https://www.image-net.org/index.php">ImageNet-1k</a> validation set.</li>
+              <li><strong>Hardware Configuration:</strong>
+                  <ul>
+                      <li>GPU: NVIDIA Tesla T4</li>
+                      <li>CPU: Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>Inference Mode Description</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>Mode</th>
+            <th>GPU Configuration </th>
+            <th>CPU Configuration </th>
+            <th>Acceleration Technology Combination</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>Normal Mode</td>
+            <td>FP32 Precision / No TRT Acceleration</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>High-Performance Mode</td>
+            <td>Optimal combination of pre-selected precision types and acceleration strategies</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.)</td>
+        </tr>
+    </tbody>
+</table>
 
 </details>
 
@@ -1699,7 +1726,7 @@ Since the general image classification pipeline includes an image classification
 <tr>
 <td>Inaccurate multi-label classification</td>
 <td>Multi-label classification module</td>
-<td><a href="../../../module_usage/tutorials/cv_modules/image_classification.en.md">Link</a></td>
+<td><a href="https://paddlepaddle.github.io/PaddleX/latest/en/module_usage/tutorials/cv_modules/image_classification.html">Link</a></td>
 </tr>
 </tbody>
 </table>

+ 46 - 16
docs/pipeline_usage/tutorials/cv_pipelines/image_classification.md

@@ -681,21 +681,51 @@ comments: true
 </tr>
 </table>
 
-**测试环境说明:**
-
-- **性能测试环境**
-  - **测试数据集**:<a href="https://www.image-net.org/index.php">ImageNet-1k</a> 验证集。
-  - **硬件配置**:
-    - GPU:NVIDIA Tesla T4
-    - CPU:Intel Xeon Gold 6271C @ 2.60GHz
-    - 其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **推理模式说明**
-
-| 模式        | GPU配置                          | CPU配置          | 加速技术组合                                |
-|-------------|----------------------------------|------------------|---------------------------------------------|
-| 常规模式    | FP32精度 / 无TRT加速             | FP32精度 / 8线程       | PaddleInference                             |
-| 高性能模式  | 选择先验精度类型和加速策略的最优组合         | FP32精度 / 8线程       | 选择先验最优后端(Paddle/OpenVINO/TRT等) |
+<strong>测试环境说明:</strong>
+
+  <ul>
+      <li><b>性能测试环境</b>
+          <ul>
+           <li><strong>测试数据集:
+             </strong>
+               <a href="https://www.image-net.org/index.php">ImageNet-1k</a> 验证集。
+             </li>
+              <li><strong>硬件配置:</strong>
+                  <ul>
+                      <li>GPU:NVIDIA Tesla T4</li>
+                      <li>CPU:Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>推理模式说明</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>模式</th>
+            <th>GPU配置</th>
+            <th>CPU配置</th>
+            <th>加速技术组合</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>常规模式</td>
+            <td>FP32精度 / 无TRT加速</td>
+            <td>FP32精度 / 8线程</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>高性能模式</td>
+            <td>选择先验精度类型和加速策略的最优组合</td>
+            <td>FP32精度 / 8线程</td>
+            <td>选择先验最优后端(Paddle/OpenVINO/TRT等)</td>
+        </tr>
+    </tbody>
+</table>
 
 </details>
 
@@ -1511,7 +1541,7 @@ print_r($result["categories"]);
 <tr>
 <td>多标签分类效果不准</td>
 <td>多标签分类模块</td>
-<td><a href="../../../module_usage/tutorials/cv_modules/image_classification.md">链接</a></td>
+<td><a href="https://paddlepaddle.github.io/PaddleX/latest/module_usage/tutorials/cv_modules/image_classification.html">链接</a></td>
 </tr>
 </tbody>
 </table>

+ 43 - 16
docs/pipeline_usage/tutorials/cv_pipelines/image_multi_label_classification.en.md

@@ -53,21 +53,48 @@ Image multi-label classification is a technique that assigns multiple relevant c
 </tbody>
 </table>
 
-**Test Environment Description**:
-
-- **Performance Test Environment**
-  - **Test Dataset**: multi-label classification task on <b><a href="https://cocodataset.org/#home">COCO2017</a></b>.
-  - **Hardware Configuration**:
-    - GPU: NVIDIA Tesla T4
-    - CPU: Intel Xeon Gold 6271C @ 2.60GHz
-    - Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **Inference Mode Description**
-
-| Mode        | GPU Configuration                        | CPU Configuration | Acceleration Technology Combination                   |
-|-------------|----------------------------------------|-------------------|---------------------------------------------------|
-| Normal Mode | FP32 Precision / No TRT Acceleration   | FP32 Precision / 8 Threads | PaddleInference                                 |
-| High-Performance Mode | Optimal combination of pre-selected precision types and acceleration strategies | FP32 Precision / 8 Threads | Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.) |
+<strong>Test Environment Description:</strong>
+
+  <ul>
+      <li><b>Performance Test Environment</b>
+          <ul>
+             <li><strong>Test Dataset:</strong>multi-label classification task on <b><a href="https://cocodataset.org/#home">COCO2017</a></b>.</li>
+              <li><strong>Hardware Configuration:</strong>
+                  <ul>
+                      <li>GPU: NVIDIA Tesla T4</li>
+                      <li>CPU: Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>Inference Mode Description</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>Mode</th>
+            <th>GPU Configuration </th>
+            <th>CPU Configuration </th>
+            <th>Acceleration Technology Combination</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>Normal Mode</td>
+            <td>FP32 Precision / No TRT Acceleration</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>High-Performance Mode</td>
+            <td>Optimal combination of pre-selected precision types and acceleration strategies</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.)</td>
+        </tr>
+    </tbody>
+</table>
 
 ## 2. Quick Start
 All model pipelines provided by PaddleX can be quickly experienced. You can experience the effect of the image multi-label classification pipeline on the community platform, or you can use the command line or Python locally to experience the effect of the image multi-label classification pipeline.
@@ -903,7 +930,7 @@ Since the general image multi-label classification pipeline includes an image mu
     <tr>
       <td>Multi-label classification is inaccurate</td>
       <td>Multi-label classification module</td>
-      <td><a href="../../../module_usage/tutorials/cv_modules/image_multilabel_classification.en.md">Link</a></td>
+      <td><a href="https://paddlepaddle.github.io/PaddleX/latest/en/module_usage/tutorials/cv_modules/image_multilabel_classification.html">Link</a></td>
     </tr>
   </tbody>
 </table>

+ 46 - 16
docs/pipeline_usage/tutorials/cv_pipelines/image_multi_label_classification.md

@@ -53,21 +53,51 @@ comments: true
 </tbody>
 </table>
 
-**测试环境说明:**
-
-- **性能测试环境**
-  - **测试数据集**:<a href="https://cocodataset.org/#home">COCO2017</a><b> 的多标签分类任务</b>
-  - **硬件配置**:
-    - GPU:NVIDIA Tesla T4
-    - CPU:Intel Xeon Gold 6271C @ 2.60GHz
-    - 其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **推理模式说明**
-
-| 模式        | GPU配置                          | CPU配置          | 加速技术组合                                |
-|-------------|----------------------------------|------------------|---------------------------------------------|
-| 常规模式    | FP32精度 / 无TRT加速             | FP32精度 / 8线程       | PaddleInference                             |
-| 高性能模式  | 选择先验精度类型和加速策略的最优组合         | FP32精度 / 8线程       | 选择先验最优后端(Paddle/OpenVINO/TRT等) |
+<strong>测试环境说明:</strong>
+
+  <ul>
+      <li><b>性能测试环境</b>
+          <ul>
+            <li><strong>测试数据集:
+             </strong>
+               <a href="https://cocodataset.org/#home">COCO2017</a><b> 的多标签分类任务</b>
+             </li>
+              <li><strong>硬件配置:</strong>
+                  <ul>
+                      <li>GPU:NVIDIA Tesla T4</li>
+                      <li>CPU:Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>推理模式说明</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>模式</th>
+            <th>GPU配置</th>
+            <th>CPU配置</th>
+            <th>加速技术组合</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>常规模式</td>
+            <td>FP32精度 / 无TRT加速</td>
+            <td>FP32精度 / 8线程</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>高性能模式</td>
+            <td>选择先验精度类型和加速策略的最优组合</td>
+            <td>FP32精度 / 8线程</td>
+            <td>选择先验最优后端(Paddle/OpenVINO/TRT等)</td>
+        </tr>
+    </tbody>
+</table>
 
 
 ## 2. 快速开始
@@ -898,7 +928,7 @@ print_r($result[&quot;categories&quot;]);
     <tr>
       <td>多标签分类效果不准</td>
       <td>多标签分类模块</td>
-      <td><a href="../../../module_usage/tutorials/cv_modules/image_multilabel_classification.md">链接</a></td>
+      <td><a href="https://paddlepaddle.github.io/PaddleX/latest/module_usage/tutorials/cv_modules/image_multilabel_classification.html">链接</a></td>
     </tr>
   </tbody>
 </table>

+ 43 - 16
docs/pipeline_usage/tutorials/cv_pipelines/instance_segmentation.en.md

@@ -160,21 +160,48 @@ Instance segmentation is a computer vision task that not only identifies the obj
 </tr>
 </table>
 
-**Test Environment Description**:
-
-- **Performance Test Environment**
-  - **Test Dataset**: <a href="https://cocodataset.org/#home">COCO2017</a> validation set.
-  - **Hardware Configuration**:
-    - GPU: NVIDIA Tesla T4
-    - CPU: Intel Xeon Gold 6271C @ 2.60GHz
-    - Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **Inference Mode Description**
-
-| Mode        | GPU Configuration                        | CPU Configuration | Acceleration Technology Combination                   |
-|-------------|----------------------------------------|-------------------|---------------------------------------------------|
-| Normal Mode | FP32 Precision / No TRT Acceleration   | FP32 Precision / 8 Threads | PaddleInference                                 |
-| High-Performance Mode | Optimal combination of pre-selected precision types and acceleration strategies | FP32 Precision / 8 Threads | Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.) |
+<strong>Test Environment Description:</strong>
+
+  <ul>
+      <li><b>Performance Test Environment</b>
+          <ul>
+           <li><strong>Test Dataset:</strong><a href="https://cocodataset.org/#home">COCO2017</a> validation set.</li>
+              <li><strong>Hardware Configuration:</strong>
+                  <ul>
+                      <li>GPU: NVIDIA Tesla T4</li>
+                      <li>CPU: Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>Inference Mode Description</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>Mode</th>
+            <th>GPU Configuration </th>
+            <th>CPU Configuration </th>
+            <th>Acceleration Technology Combination</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>Normal Mode</td>
+            <td>FP32 Precision / No TRT Acceleration</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>High-Performance Mode</td>
+            <td>Optimal combination of pre-selected precision types and acceleration strategies</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.)</td>
+        </tr>
+    </tbody>
+</table>
 
 </details>
 
@@ -1048,7 +1075,7 @@ Since the general instance segmentation pipeline includes an instance segmentati
 <tr>
 <td>Prediction results are not as expected</td>
 <td>Instance Segmentation Module</td>
-<td><a href="../../../module_usage/tutorials/cv_modules/instance_segmentation.en.md">Link</a></td>
+<td><a href="https://paddlepaddle.github.io/PaddleX/latest/en/module_usage/tutorials/cv_modules/instance_segmentation.html">Link</a></td>
 </tr>
 </tbody>
 </table>

+ 46 - 16
docs/pipeline_usage/tutorials/cv_pipelines/instance_segmentation.md

@@ -163,21 +163,51 @@ comments: true
 </tr>
 </table>
 
-**测试环境说明:**
-
-- **性能测试环境**
-  - **测试数据集**:<a href="https://cocodataset.org/#home">COCO2017</a>验证集。
-  - **硬件配置**:
-    - GPU:NVIDIA Tesla T4
-    - CPU:Intel Xeon Gold 6271C @ 2.60GHz
-    - 其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **推理模式说明**
-
-| 模式        | GPU配置                          | CPU配置          | 加速技术组合                                |
-|-------------|----------------------------------|------------------|---------------------------------------------|
-| 常规模式    | FP32精度 / 无TRT加速             | FP32精度 / 8线程       | PaddleInference                             |
-| 高性能模式  | 选择先验精度类型和加速策略的最优组合         | FP32精度 / 8线程       | 选择先验最优后端(Paddle/OpenVINO/TRT等) |
+<strong>测试环境说明:</strong>
+
+  <ul>
+      <li><b>性能测试环境</b>
+          <ul>
+            <li><strong>测试数据集:
+             </strong>
+               <a href="https://cocodataset.org/#home">COCO2017</a>验证集。
+             </li>
+              <li><strong>硬件配置:</strong>
+                  <ul>
+                      <li>GPU:NVIDIA Tesla T4</li>
+                      <li>CPU:Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>推理模式说明</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>模式</th>
+            <th>GPU配置</th>
+            <th>CPU配置</th>
+            <th>加速技术组合</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>常规模式</td>
+            <td>FP32精度 / 无TRT加速</td>
+            <td>FP32精度 / 8线程</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>高性能模式</td>
+            <td>选择先验精度类型和加速策略的最优组合</td>
+            <td>FP32精度 / 8线程</td>
+            <td>选择先验最优后端(Paddle/OpenVINO/TRT等)</td>
+        </tr>
+    </tbody>
+</table>
 
 </details>
 
@@ -1046,7 +1076,7 @@ print_r($result["instances"]);
 <tr>
 <td>预测结果不达预期</td>
 <td>实例分割模块</td>
-<td><a href="../../../module_usage/tutorials/cv_modules/instance_segmentation.md">链接</a></td>
+<td><a href="https://paddlepaddle.github.io/PaddleX/latest/module_usage/tutorials/cv_modules/instance_segmentation.html">链接</a></td>
 </tr>
 </tbody>
 </table>

+ 43 - 16
docs/pipeline_usage/tutorials/cv_pipelines/object_detection.en.md

@@ -347,21 +347,48 @@ Object detection aims to identify the categories and locations of multiple objec
 </tr>
 </table>
 
-**Test Environment Description**:
-
-- **Performance Test Environment**
-  - **Test Dataset**: <a href="https://cocodataset.org/#home">COCO2017</a> validation set.
-  - **Hardware Configuration**:
-    - GPU: NVIDIA Tesla T4
-    - CPU: Intel Xeon Gold 6271C @ 2.60GHz
-    - Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **Inference Mode Description**
-
-| Mode        | GPU Configuration                        | CPU Configuration | Acceleration Technology Combination                   |
-|-------------|----------------------------------------|-------------------|---------------------------------------------------|
-| Normal Mode | FP32 Precision / No TRT Acceleration   | FP32 Precision / 8 Threads | PaddleInference                                 |
-| High-Performance Mode | Optimal combination of pre-selected precision types and acceleration strategies | FP32 Precision / 8 Threads | Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.) |
+<strong>Test Environment Description:</strong>
+
+  <ul>
+      <li><b>Performance Test Environment</b>
+          <ul>
+               <li><strong>Test Dataset:</strong><a href="https://cocodataset.org/#home">COCO2017</a> validation set.</li>
+              <li><strong>Hardware Configuration:</strong>
+                  <ul>
+                      <li>GPU: NVIDIA Tesla T4</li>
+                      <li>CPU: Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>Inference Mode Description</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>Mode</th>
+            <th>GPU Configuration </th>
+            <th>CPU Configuration </th>
+            <th>Acceleration Technology Combination</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>Normal Mode</td>
+            <td>FP32 Precision / No TRT Acceleration</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>High-Performance Mode</td>
+            <td>Optimal combination of pre-selected precision types and acceleration strategies</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.)</td>
+        </tr>
+    </tbody>
+</table>
 
 </details>
 
@@ -1236,7 +1263,7 @@ You can choose the appropriate method to deploy the model pipeline based on your
 If the default model weights provided by the general object detection pipeline do not meet your accuracy or speed requirements in your scenario, you can try further <b>fine-tuning</b> the existing model using <b>your own specific domain or application scenario data</b> to improve the recognition performance of the general object detection pipeline in your scenario.
 
 ### 4.1 Model Fine-Tuning
-Since the general object detection pipeline includes an object detection module, if the performance of the model pipeline is not as expected, you need to refer to the [Custom Development](../../../module_usage/tutorials/cv_modules/object_detection.en.md#iv-custom-development) section in the [Object Detection Module Development Tutorial](../../../module_usage/tutorials/cv_modules/object_detection.en.md) to fine-tune the object detection model using your private dataset.
+Since the general object detection pipeline includes an object detection module, if the performance of the model pipeline is not as expected, you need to refer to the <b>Custom Development</b> section in the [Object Detection Module Development Tutorial](https://paddlepaddle.github.io/PaddleX/latest/en/module_usage/tutorials/cv_modules/object_detection.html) to fine-tune the object detection model using your private dataset.
 
 ### 4.2 Model Application
 After completing the fine-tuning training with your private dataset, you will obtain a local model weight file.

+ 46 - 16
docs/pipeline_usage/tutorials/cv_pipelines/object_detection.md

@@ -364,21 +364,51 @@ comments: true
 </tr>
 </table>
 
-**测试环境说明:**
-
-- **性能测试环境**
-  - **测试数据集**:<a href="https://cocodataset.org/#home">COCO2017</a>验证集。
-  - **硬件配置**:
-    - GPU:NVIDIA Tesla T4
-    - CPU:Intel Xeon Gold 6271C @ 2.60GHz
-    - 其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **推理模式说明**
-
-| 模式        | GPU配置                          | CPU配置          | 加速技术组合                                |
-|-------------|----------------------------------|------------------|---------------------------------------------|
-| 常规模式    | FP32精度 / 无TRT加速             | FP32精度 / 8线程       | PaddleInference                             |
-| 高性能模式  | 选择先验精度类型和加速策略的最优组合         | FP32精度 / 8线程       | 选择先验最优后端(Paddle/OpenVINO/TRT等) |
+<strong>测试环境说明:</strong>
+
+  <ul>
+      <li><b>性能测试环境</b>
+          <ul>
+            <li><strong>测试数据集:
+             </strong>
+               <a href="https://cocodataset.org/#home">COCO2017</a>验证集。
+             </li>
+              <li><strong>硬件配置:</strong>
+                  <ul>
+                      <li>GPU:NVIDIA Tesla T4</li>
+                      <li>CPU:Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>推理模式说明</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>模式</th>
+            <th>GPU配置</th>
+            <th>CPU配置</th>
+            <th>加速技术组合</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>常规模式</td>
+            <td>FP32精度 / 无TRT加速</td>
+            <td>FP32精度 / 8线程</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>高性能模式</td>
+            <td>选择先验精度类型和加速策略的最优组合</td>
+            <td>FP32精度 / 8线程</td>
+            <td>选择先验最优后端(Paddle/OpenVINO/TRT等)</td>
+        </tr>
+    </tbody>
+</table>
 
 </details>
 
@@ -1186,7 +1216,7 @@ print_r($result["detectedObjects"]);
 如果通用目标检测产线提供的默认模型权重在您的场景中,精度或速度不满意,您可以尝试利用<b>您自己拥有的特定领域或应用场景的数据</b>对现有模型进行进一步的<b>微调</b>,以提升通用目标检测产线的在您的场景中的识别效果。
 
 ### 4.1 模型微调
-由于通用目标检测产线包含目标检测模块,如果模型产线的效果不及预期,那么您需要参考[目标检测模块开发教程](../../../module_usage/tutorials/cv_modules/object_detection.md)中的[二次开发](../../../module_usage/tutorials/cv_modules/object_detection.md#四二次开发)章节,使用您的私有数据集对目标检测模型进行微调。
+由于通用目标检测产线包含目标检测模块,如果模型产线的效果不及预期,那么您需要参考[目标检测模块开发教程](https://paddlepaddle.github.io/PaddleX/latest/module_usage/tutorials/cv_modules/object_detection.html)中的<b>二次开发</b>章节,使用您的私有数据集对目标检测模型进行微调。
 
 ### 4.2 模型应用
 当您使用私有数据集完成微调训练后,可获得本地模型权重文件。

+ 42 - 15
docs/pipeline_usage/tutorials/cv_pipelines/open_vocabulary_detection.en.md

@@ -35,21 +35,48 @@ Open vocabulary object detection is an advanced object detection technology that
 <td rowspan="3">An open vocabulary object detection model trained on O365, GoldG, and Cap4M datasets. The text encoder uses Bert, and the visual model part adopts DINO overall, with additional cross-modal fusion modules designed, achieving good results in the field of open vocabulary object detection.</td>
 </tr>
 </table>
-**Test Environment Description**:
-
-- **Performance Test Environment**
-  - **Test Dataset**: COCO val2017 validation set
-  - **Hardware Configuration**:
-    - GPU: NVIDIA Tesla T4
-    - CPU: Intel Xeon Gold 6271C @ 2.60GHz
-    - Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **Inference Mode Description**
-
-| Mode        | GPU Configuration                        | CPU Configuration | Acceleration Technology Combination                   |
-|-------------|----------------------------------------|-------------------|---------------------------------------------------|
-| Normal Mode | FP32 Precision / No TRT Acceleration   | FP32 Precision / 8 Threads | PaddleInference                                 |
-| High-Performance Mode | Optimal combination of pre-selected precision types and acceleration strategies | FP32 Precision / 8 Threads | Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.) |
+<strong>Test Environment Description:</strong>
+
+  <ul>
+      <li><b>Performance Test Environment</b>
+          <ul>
+            <li><strong>Test Dataset:</strong>COCO val2017 validation set</li>
+              <li><strong>Hardware Configuration:</strong>
+                  <ul>
+                      <li>GPU: NVIDIA Tesla T4</li>
+                      <li>CPU: Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>Inference Mode Description</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>Mode</th>
+            <th>GPU Configuration </th>
+            <th>CPU Configuration </th>
+            <th>Acceleration Technology Combination</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>Normal Mode</td>
+            <td>FP32 Precision / No TRT Acceleration</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>High-Performance Mode</td>
+            <td>Optimal combination of pre-selected precision types and acceleration strategies</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.)</td>
+        </tr>
+    </tbody>
+</table>
 
 ## 2. Quick Start
 

+ 45 - 15
docs/pipeline_usage/tutorials/cv_pipelines/open_vocabulary_detection.md

@@ -36,21 +36,51 @@ comments: true
 </tr>
 </table>
 
-**测试环境说明:**
-
-- **性能测试环境**
-  - **测试数据集**:COCO val2017 验证集
-  - **硬件配置**:
-    - GPU:NVIDIA Tesla T4
-    - CPU:Intel Xeon Gold 6271C @ 2.60GHz
-    - 其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **推理模式说明**
-
-| 模式        | GPU配置                          | CPU配置          | 加速技术组合                                |
-|-------------|----------------------------------|------------------|---------------------------------------------|
-| 常规模式    | FP32精度 / 无TRT加速             | FP32精度 / 8线程       | PaddleInference                             |
-| 高性能模式  | 选择先验精度类型和加速策略的最优组合         | FP32精度 / 8线程       | 选择先验最优后端(Paddle/OpenVINO/TRT等) |
+<strong>测试环境说明:</strong>
+
+  <ul>
+      <li><b>性能测试环境</b>
+          <ul>
+          <li><strong>测试数据集:
+             </strong>
+               COCO val2017 验证集
+             </li>
+              <li><strong>硬件配置:</strong>
+                  <ul>
+                      <li>GPU:NVIDIA Tesla T4</li>
+                      <li>CPU:Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>推理模式说明</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>模式</th>
+            <th>GPU配置</th>
+            <th>CPU配置</th>
+            <th>加速技术组合</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>常规模式</td>
+            <td>FP32精度 / 无TRT加速</td>
+            <td>FP32精度 / 8线程</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>高性能模式</td>
+            <td>选择先验精度类型和加速策略的最优组合</td>
+            <td>FP32精度 / 8线程</td>
+            <td>选择先验最优后端(Paddle/OpenVINO/TRT等)</td>
+        </tr>
+    </tbody>
+</table>
 
 ## 2. 快速开始
 

+ 41 - 14
docs/pipeline_usage/tutorials/cv_pipelines/open_vocabulary_segmentation.en.md

@@ -38,20 +38,47 @@ Open vocabulary segmentation is an image segmentation task that aims to segment
 </tr>
 </table>
 
-**Test Environment Description**:
-
-- **Performance Test Environment**
-  - **Hardware Configuration**:
-    - GPU: NVIDIA Tesla T4
-    - CPU: Intel Xeon Gold 6271C @ 2.60GHz
-    - Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **Inference Mode Description**
-
-| Mode        | GPU Configuration                        | CPU Configuration | Acceleration Technology Combination                   |
-|-------------|----------------------------------------|-------------------|---------------------------------------------------|
-| Normal Mode | FP32 Precision / No TRT Acceleration   | FP32 Precision / 8 Threads | PaddleInference                                 |
-| High-Performance Mode | Optimal combination of pre-selected precision types and acceleration strategies | FP32 Precision / 8 Threads | Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.) |
+<strong>Test Environment Description:</strong>
+
+  <ul>
+      <li><b>Performance Test Environment</b>
+          <ul>
+              <li><strong>Hardware Configuration:</strong>
+                  <ul>
+                      <li>GPU: NVIDIA Tesla T4</li>
+                      <li>CPU: Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>Inference Mode Description</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>Mode</th>
+            <th>GPU Configuration </th>
+            <th>CPU Configuration </th>
+            <th>Acceleration Technology Combination</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>Normal Mode</td>
+            <td>FP32 Precision / No TRT Acceleration</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>High-Performance Mode</td>
+            <td>Optimal combination of pre-selected precision types and acceleration strategies</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.)</td>
+        </tr>
+    </tbody>
+</table>
 
 ## 2. Quick Start
 

+ 41 - 14
docs/pipeline_usage/tutorials/cv_pipelines/open_vocabulary_segmentation.md

@@ -38,20 +38,47 @@ comments: true
 </tr>
 </table>
 
-**测试环境说明:**
-
-- **性能测试环境**
-  - **硬件配置**:
-    - GPU:NVIDIA Tesla T4
-    - CPU:Intel Xeon Gold 6271C @ 2.60GHz
-    - 其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **推理模式说明**
-
-| 模式        | GPU配置                          | CPU配置          | 加速技术组合                                |
-|-------------|----------------------------------|------------------|---------------------------------------------|
-| 常规模式    | FP32精度 / 无TRT加速             | FP32精度 / 8线程       | PaddleInference                             |
-| 高性能模式  | 选择先验精度类型和加速策略的最优组合         | FP32精度 / 8线程       | 选择先验最优后端(Paddle/OpenVINO/TRT等) |
+<strong>测试环境说明:</strong>
+
+  <ul>
+      <li><b>性能测试环境</b>
+          <ul>
+              <li><strong>硬件配置:</strong>
+                  <ul>
+                      <li>GPU:NVIDIA Tesla T4</li>
+                      <li>CPU:Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>推理模式说明</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>模式</th>
+            <th>GPU配置</th>
+            <th>CPU配置</th>
+            <th>加速技术组合</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>常规模式</td>
+            <td>FP32精度 / 无TRT加速</td>
+            <td>FP32精度 / 8线程</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>高性能模式</td>
+            <td>选择先验精度类型和加速策略的最优组合</td>
+            <td>FP32精度 / 8线程</td>
+            <td>选择先验最优后端(Paddle/OpenVINO/TRT等)</td>
+        </tr>
+    </tbody>
+</table>
 
 ## 2. 快速开始
 

+ 49 - 19
docs/pipeline_usage/tutorials/cv_pipelines/pedestrian_attribute_recognition.en.md

@@ -64,23 +64,53 @@ Pedestrian attribute recognition is a key function in computer vision systems, u
 </tbody>
 </table>
 
-**Test Environment Description**:
-
-- **Performance Test Environment**
-  - **Test Dataset**:
-    - Pedestrian Detection Model: CrowdHuman Dataset.
-    - Pedestrian Attribute Recognition Model: PaddleX Internal Self-built Dataset.
-  - **Hardware Configuration**:
-    - GPU: NVIDIA Tesla T4
-    - CPU: Intel Xeon Gold 6271C @ 2.60GHz
-    - Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **Inference Mode Description**
-
-| Mode        | GPU Configuration                        | CPU Configuration | Acceleration Technology Combination                   |
-|-------------|----------------------------------------|-------------------|---------------------------------------------------|
-| Normal Mode | FP32 Precision / No TRT Acceleration   | FP32 Precision / 8 Threads | PaddleInference                                 |
-| High-Performance Mode | Optimal combination of pre-selected precision types and acceleration strategies | FP32 Precision / 8 Threads | Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.) |
+<strong>Test Environment Description:</strong>
+
+  <ul>
+      <li><b>Performance Test Environment</b>
+          <ul>
+               <li><strong>Test Dataset:</strong>
+                  <ul>
+                   <li>Pedestrian Detection Model: CrowdHuman Dataset.</li>
+                   <li>Pedestrian Attribute Recognition Model: PaddleX Internal Self-built Dataset.</li>
+                  </ul>
+               </li>
+              <li><strong>Hardware Configuration:</strong>
+                  <ul>
+                      <li>GPU: NVIDIA Tesla T4</li>
+                      <li>CPU: Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>Inference Mode Description</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>Mode</th>
+            <th>GPU Configuration </th>
+            <th>CPU Configuration </th>
+            <th>Acceleration Technology Combination</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>Normal Mode</td>
+            <td>FP32 Precision / No TRT Acceleration</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>High-Performance Mode</td>
+            <td>Optimal combination of pre-selected precision types and acceleration strategies</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.)</td>
+        </tr>
+    </tbody>
+</table>
 
 ## 2. Quick Start
 
@@ -616,12 +646,12 @@ Since the pedestrian attribute recognition pipeline includes both a pedestrian a
 <tr>
 <td>Inaccurate pedestrian detection</td>
 <td>Pedestrian Detection Module</td>
-<td><a href="../../../module_usage/tutorials/cv_modules/human_detection.en.md">Link</a></td>
+<td><a href="https://paddlepaddle.github.io/PaddleX/latest/en/module_usage/tutorials/cv_modules/human_detection.html">Link</a></td>
 </tr>
 <tr>
 <td>Inaccurate attribute recognition</td>
 <td>Pedestrian Attribute Recognition Module</td>
-<td><a href="../../../module_usage/tutorials/cv_modules/pedestrian_attribute_recognition.en.md">Link</a></td>
+<td><a href="https://paddlepaddle.github.io/PaddleX/latest/en/module_usage/tutorials/cv_modules/pedestrian_attribute_recognition.html">Link</a></td>
 </tr>
 </tbody>
 </table>

+ 50 - 19
docs/pipeline_usage/tutorials/cv_pipelines/pedestrian_attribute_recognition.md

@@ -65,23 +65,54 @@ comments: true
 </tbody>
 </table>
 
-**测试环境说明:**
-
-- **性能测试环境**
-  - **测试数据集**
-    - 行人检测模型:CrowdHuman数据集。
-    - 行人属性是别模型:PaddleX 内部自建数据集。
-  - **硬件配置**:
-    - GPU:NVIDIA Tesla T4
-    - CPU:Intel Xeon Gold 6271C @ 2.60GHz
-    - 其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **推理模式说明**
-
-| 模式        | GPU配置                          | CPU配置          | 加速技术组合                                |
-|-------------|----------------------------------|------------------|---------------------------------------------|
-| 常规模式    | FP32精度 / 无TRT加速             | FP32精度 / 8线程       | PaddleInference                             |
-| 高性能模式  | 选择先验精度类型和加速策略的最优组合         | FP32精度 / 8线程       | 选择先验最优后端(Paddle/OpenVINO/TRT等) |
+<strong>测试环境说明:</strong>
+
+  <ul>
+      <li><b>性能测试环境</b>
+          <ul>
+             <li><strong>测试数据集:
+             </strong>
+                <ul>
+                  <li>行人检测模型:CrowdHuman数据集。</li>
+                  <li> 行人属性识别模型:PaddleX 内部自建数据集。</li>
+                </ul>
+             </li>
+              <li><strong>硬件配置:</strong>
+                  <ul>
+                      <li>GPU:NVIDIA Tesla T4</li>
+                      <li>CPU:Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>推理模式说明</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>模式</th>
+            <th>GPU配置</th>
+            <th>CPU配置</th>
+            <th>加速技术组合</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>常规模式</td>
+            <td>FP32精度 / 无TRT加速</td>
+            <td>FP32精度 / 8线程</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>高性能模式</td>
+            <td>选择先验精度类型和加速策略的最优组合</td>
+            <td>FP32精度 / 8线程</td>
+            <td>选择先验最优后端(Paddle/OpenVINO/TRT等)</td>
+        </tr>
+    </tbody>
+</table>
 
 ## 2. 快速开始
 PaddleX 所提供的模型产线均可以快速体验效果,你可以在星河社区线体验行人属性识别产线的效果,也可以在本地使用命令行或 Python 体验行人属性识别产线的效果。
@@ -612,12 +643,12 @@ print(result["pedestrians"])
 <tr>
 <td>行人检测不准</td>
 <td>行人检测模块</td>
-<td><a href="../../../module_usage/tutorials/cv_modules/human_detection.md">链接</a></td>
+<td><a href="https://paddlepaddle.github.io/PaddleX/latest/module_usage/tutorials/cv_modules/human_detection.html">链接</a></td>
 </tr>
 <tr>
 <td>属性识别不准</td>
 <td>行人属性识别模块</td>
-<td><a href="../../../module_usage/tutorials/cv_modules/pedestrian_attribute_recognition.md">链接</a></td>
+<td><a href="https://paddlepaddle.github.io/PaddleX/latest/module_usage/tutorials/cv_modules/pedestrian_attribute_recognition.html">链接</a></td>
 </tr>
 </tbody>
 </table>

+ 43 - 83
docs/pipeline_usage/tutorials/cv_pipelines/rotated_object_detection.en.md

@@ -34,21 +34,48 @@ Rotated object detection is a variant of the object detection module, specifical
 </tr>
 </table>
 
-**Test Environment Description**:
-
-- **Performance Test Environment**
-  - **Test Dataset**: <a href="https://captain-whu.github.io/DOTA/">DOTA</a> validation set.
-  - **Hardware Configuration**:
-    - GPU: NVIDIA Tesla T4
-    - CPU: Intel Xeon Gold 6271C @ 2.60GHz
-    - Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **Inference Mode Description**
-
-| Mode        | GPU Configuration                        | CPU Configuration | Acceleration Technology Combination                   |
-|-------------|----------------------------------------|-------------------|---------------------------------------------------|
-| Normal Mode | FP32 Precision / No TRT Acceleration   | FP32 Precision / 8 Threads | PaddleInference                                 |
-| High-Performance Mode | Optimal combination of pre-selected precision types and acceleration strategies | FP32 Precision / 8 Threads | Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.) |
+<strong>Test Environment Description:</strong>
+
+  <ul>
+      <li><b>Performance Test Environment</b>
+          <ul>
+               <li><strong>Test Dataset:</strong> <a href="https://captain-whu.github.io/DOTA/">DOTA</a> validation set.</li>
+              <li><strong>Hardware Configuration:</strong>
+                  <ul>
+                      <li>GPU: NVIDIA Tesla T4</li>
+                      <li>CPU: Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>Other Environments: Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>Inference Mode Description</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>Mode</th>
+            <th>GPU Configuration </th>
+            <th>CPU Configuration </th>
+            <th>Acceleration Technology Combination</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>Normal Mode</td>
+            <td>FP32 Precision / No TRT Acceleration</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>High-Performance Mode</td>
+            <td>Optimal combination of pre-selected precision types and acceleration strategies</td>
+            <td>FP32 Precision / 8 Threads</td>
+            <td>Pre-selected optimal backend (Paddle/OpenVINO/TRT, etc.)</td>
+        </tr>
+    </tbody>
+</table>
 
 ## 2. Quick Start
 
@@ -899,73 +926,6 @@ print_r($result["detectedObjects"]);
 📱 <b>Edge Deployment</b>: Edge deployment is a method of placing computing and data processing capabilities directly on user devices, allowing them to process data locally without relying on remote servers. PaddleX supports deploying models on edge devices such as Android. For detailed instructions, please refer to the [PaddleX Edge Deployment Guide](../../../pipeline_deploy/edge_deploy.en.md).
 You can choose the appropriate deployment method based on your needs to integrate the model pipeline into subsequent AI applications.
 
-
-## 4. Custom Development
-If the default model weights provided by the Rotated Object Detection Pipeline do not meet your requirements in terms of accuracy or speed, you can attempt to <b>fine-tune</b> the existing models using <b>your own domain-specific or application-specific data</b> to improve the detection performance in your scenario.
-
-### 4.1 Model Fine-Tuning
-Since the Rotated Object Detection Pipeline includes a rotated object detection module, if the pipeline's performance is not satisfactory, you can analyze the poorly detected images and refer to the fine-tuning tutorial links in the table below for model fine-tuning.
-
-
-<table>
-  <thead>
-    <tr>
-      <th>Scenario</th>
-      <th>Fine-Tuning Module</th>
-      <th>Fine-Tuning Reference Link</th>
-    </tr>
-  </thead>
-  <tbody>
-    <tr>
-      <td>Prediction results are not satisfactory</td>
-      <td>Rotated Object Detection Module</td>
-      <td><a href="../../../module_usage/tutorials/cv_modules/rotated_object_detection.en.md">Link</a></td>
-    </tr>
-  </tbody>
-</table>
-
-### 4.2 Model Application
-After fine-tuning with your private dataset, you will obtain the local model weight file.
-
-To use the fine-tuned model weights, simply modify the pipeline configuration file by replacing the path of the fine-tuned model weights with the corresponding location in the pipeline configuration file:
-
-<details><summary>PHP</summary>
-
-<pre><code class="language-php">&lt;?php
-
-$API_URL = "http://localhost:8080/small-object-detection"; // Service URL
-$image_path = "./demo.jpg";
-$output_image_path = "./out.jpg";
-
-// Encode the local image in Base64
-$image_data = base64_encode(file_get_contents($image_path));
-$payload = array("image" => $image_data); // Base64-encoded file content or image URL
-
-// Call the API
-$ch = curl_init($API_URL);
-curl_setopt($ch, CURLOPT_POST, true);
-curl_setopt($ch, CURLOPT_POSTFIELDS, json_encode($payload));
-curl_setopt($ch, CURLOPT_HTTPHEADER, array('Content-Type: application/json'));
-curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
-$response = curl_exec($ch);
-curl_close($ch);
-
-// Process the response data from the API
-$result = json_decode($response, true)["result"];
-file_put_contents($output_image_path, base64_decode($result["image"]));
-echo "Output image saved at " . $output_image_path . "\n";
-echo "\nDetected objects:\n";
-print_r($result["detectedObjects"]);
-
-?&gt;
-</code></pre></details>
-</details>
-<br/>
-
-📱 <b>Edge Deployment</b>: Edge deployment is a method of placing computing and data processing capabilities directly on user devices, allowing them to process data locally without relying on remote servers. PaddleX supports deploying models on edge devices such as Android. For detailed instructions, please refer to the [PaddleX Edge Deployment Guide](../../../pipeline_deploy/edge_deploy.en.md).
-You can choose the appropriate deployment method based on your needs to integrate the model pipeline into subsequent AI applications.
-
-
 ## 4. Custom Development
 If the default model weights provided by the Rotated Object Detection Pipeline do not meet your requirements in terms of accuracy or speed, you can attempt to <b>fine-tune</b> the existing models using <b>your own domain-specific or application-specific data</b> to improve the detection performance in your scenario.
 
@@ -985,7 +945,7 @@ Since the Rotated Object Detection Pipeline includes a rotated object detection
     <tr>
       <td>Prediction results are not satisfactory</td>
       <td>Rotated Object Detection Module</td>
-      <td><a href="../../../module_usage/tutorials/cv_modules/rotated_object_detection.en.md">Link</a></td>
+      <td><a href="https://paddlepaddle.github.io/PaddleX/latest/en/module_usage/tutorials/cv_modules/rotated_object_detection.html">Link</a></td>
     </tr>
   </tbody>
 </table>

+ 46 - 16
docs/pipeline_usage/tutorials/cv_pipelines/rotated_object_detection.md

@@ -34,21 +34,51 @@ comments: true
 </tr>
 </table>
 
-**测试环境说明:**
-
-- **性能测试环境**
-  - **测试数据集**:<a href="https://captain-whu.github.io/DOTA/">DOTA</a>验证集。
-  - **硬件配置**:
-    - GPU:NVIDIA Tesla T4
-    - CPU:Intel Xeon Gold 6271C @ 2.60GHz
-    - 其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2
-
-- **推理模式说明**
-
-| 模式        | GPU配置                          | CPU配置          | 加速技术组合                                |
-|-------------|----------------------------------|------------------|---------------------------------------------|
-| 常规模式    | FP32精度 / 无TRT加速             | FP32精度 / 8线程       | PaddleInference                             |
-| 高性能模式  | 选择先验精度类型和加速策略的最优组合         | FP32精度 / 8线程       | 选择先验最优后端(Paddle/OpenVINO/TRT等) |
+<strong>测试环境说明:</strong>
+
+  <ul>
+      <li><b>性能测试环境</b>
+          <ul>
+                       <li><strong>测试数据集:
+             </strong>
+                <a href="https://captain-whu.github.io/DOTA/">DOTA</a>验证集。
+             </li>
+              <li><strong>硬件配置:</strong>
+                  <ul>
+                      <li>GPU:NVIDIA Tesla T4</li>
+                      <li>CPU:Intel Xeon Gold 6271C @ 2.60GHz</li>
+                      <li>其他环境:Ubuntu 20.04 / cuDNN 8.6 / TensorRT 8.5.2.2</li>
+                  </ul>
+              </li>
+          </ul>
+      </li>
+      <li><b>推理模式说明</b></li>
+  </ul>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>模式</th>
+            <th>GPU配置</th>
+            <th>CPU配置</th>
+            <th>加速技术组合</th>
+        </tr>
+    </thead>
+    <tbody>
+        <tr>
+            <td>常规模式</td>
+            <td>FP32精度 / 无TRT加速</td>
+            <td>FP32精度 / 8线程</td>
+            <td>PaddleInference</td>
+        </tr>
+        <tr>
+            <td>高性能模式</td>
+            <td>选择先验精度类型和加速策略的最优组合</td>
+            <td>FP32精度 / 8线程</td>
+            <td>选择先验最优后端(Paddle/OpenVINO/TRT等)</td>
+        </tr>
+    </tbody>
+</table>
 
 ## 2. 快速开始
 
@@ -581,7 +611,7 @@ print(result["detectedObjects"])
     <tr>
       <td>预测结果不达预期</td>
       <td>旋转目标检测模块</td>
-      <td><a href="../../../module_usage/tutorials/cv_modules/rotated_object_detection.md">链接</a></td>
+      <td><a href="https://paddlepaddle.github.io/PaddleX/latest/module_usage/tutorials/cv_modules/rotated_object_detection.html">链接</a></td>
     </tr>
   </tbody>
 </table>

Some files were not shown because too many files changed in this diff