瀏覽代碼

Merge pull request #144 from SunAhong1993/syf0609

modify the model zoo benchmark
Jason 5 年之前
父節點
當前提交
99f52dcc2f
共有 1 個文件被更改,包括 43 次插入35 次删除
  1. 43 35
      docs/appendix/model_zoo.md

+ 43 - 35
docs/appendix/model_zoo.md

@@ -6,48 +6,56 @@
 
 | 模型  | 模型大小 | 预测速度(毫秒) | Top1准确率(%) | Top5准确率(%) |
 | :----|  :------- | :----------- | :--------- | :--------- |
-| ResNet18| 46.9MB   | -        | 71.0     | 89.9     |
-| ResNet34| 87.5MB   | -        | 74.6    | 92.1    |
-| ResNet50| 102.7MB  | -        | 76.5     | 93.0     |
-| ResNet101 |179.1MB  | -      | 77.6     | 93.6  |
-| ResNet50_vd |102.8MB  |-        | 79.1     | 94.4     |
-| ResNet101_vd| 179.2MB  | -       | 80.2   | 95.0     |
-| ResNet50_vd_ssld |102.8MB  | -        | 82.4     | 96.1     |
-| ResNet101_vd_ssld| 179.2MB  | -       | 83.7   | 96.7     |
-| DarkNet53|166.9MB  | -       | 78.0     | 94.1     |
-| MobileNetV1 | 16.0MB   | -        | 71.0     | 89.7    |
-| MobileNetV2 | 14.0MB   | -        | 72.2     | 90.7    |
-| MobileNetV3_large|  21.0MB   | -        | 75.3    | 93.2   |
-| MobileNetV3_small |  12.0MB   | -        | 68.2    | 88.1     |
-| MobileNetV3_large_ssld|  21.0MB   | -        | 79.0     | 94.5     |
-| MobileNetV3_small_ssld |  12.0MB   | -        | 71.3     | 90.1     |
-| Xception41 |92.4MB   | -       | 79.6    | 94.4     |
-| Xception65 | 144.6MB  | -       | 80.3     | 94.5     |
-| DenseNet121 | 32.8MB   | -       | 75.7     | 92.6     |
-| DenseNet161|116.3MB  | -       | 78.6     | 94.1     |
-| DenseNet201|  84.6MB   | -       | 77.6     | 93.7     |
-| ShuffleNetV2 | 9.0MB   | -        | 68.8     | 88.5     |
-| HRNet_W18 | 21.29MB | - | 76.9 | 93.4 |
+| [ResNet18](https://paddle-imagenet-models-name.bj.bcebos.com/ResNet18_pretrained.tar)| 46.2MB   | 3.72882        | 71.0     | 89.9     |
+| [ResNet34](https://paddle-imagenet-models-name.bj.bcebos.com/ResNet34_pretrained.tar)| 87.9MB   | 5.50876        | 74.6    | 92.1    |
+| [ResNet50](https://paddle-imagenet-models-name.bj.bcebos.com/ResNet50_pretrained.tar)| 103.4MB  | 7.76659       | 76.5     | 93.0     |
+| [ResNet101](https://paddle-imagenet-models-name.bj.bcebos.com/ResNet101_pretrained.tar) |180.4MB  | 13.80876      | 77.6     | 93.6  |
+| [ResNet50_vd](https://paddle-imagenet-models-name.bj.bcebos.com/ResNet50_vd_pretrained.tar) |103.5MB  | 8.20476       | 79.1     | 94.4     |
+| [ResNet101_vd](https://paddle-imagenet-models-name.bj.bcebos.com/ResNet101_vd_pretrained.tar)| 180.5MB  | 14.24643       | 80.2   | 95.0     |
+| [ResNet50_vd_ssld](https://paddle-imagenet-models-name.bj.bcebos.com/ResNet50_vd_ssld_pretrained.tar) |103.5MB  | 7.79264       | 82.4     | 96.1     |
+| [ResNet101_vd_ssld](https://paddle-imagenet-models-name.bj.bcebos.com/ResNet101_vd_ssld_pretrained.tar)| 180.5MB  | 13.34580       | 83.7   | 96.7     |
+| [DarkNet53](https://paddle-imagenet-models-name.bj.bcebos.com/DarkNet53_ImageNet1k_pretrained.tar)|167.4MB  | 8.82047       | 78.0     | 94.1     |
+| [MobileNetV1](https://paddle-imagenet-models-name.bj.bcebos.com/MobileNetV1_pretrained.tar) | 17.4MB   | 3.42838        | 71.0     | 89.7    |
+| [MobileNetV2](https://paddle-imagenet-models-name.bj.bcebos.com/MobileNetV2_pretrained.tar) | 15.0MB   | 5.92667        | 72.2     | 90.7    |
+| [MobileNetV3_large](https://paddle-imagenet-models-name.bj.bcebos.com/MobileNetV3_large_x1_0_pretrained.tar)|  22.8MB   | 8.31428        | 75.3    | 93.2   |
+| [MobileNetV3_small](https://paddle-imagenet-models-name.bj.bcebos.com/MobileNetV3_small_x1_0_pretrained.tar) |  12.5MB   | 7.30689        | 68.2    | 88.1     |
+| [MobileNetV3_large_ssld](https://paddle-imagenet-models-name.bj.bcebos.com/MobileNetV3_large_x1_0_ssld_pretrained.tar)|  22.8MB   | 8.06651        | 79.0     | 94.5     |
+| [MobileNetV3_small_ssld](https://paddle-imagenet-models-name.bj.bcebos.com/MobileNetV3_small_x1_0_ssld_pretrained.tar) |  12.5MB   | 7.08837        | 71.3     | 90.1     |
+| [Xception41](https://paddle-imagenet-models-name.bj.bcebos.com/Xception41_deeplab_pretrained.tar) | 109.2MB   | 8.15611      | 79.6    | 94.4     |
+| [Xception65](https://paddle-imagenet-models-name.bj.bcebos.com/Xception65_deeplab_pretrained.tar) | 161.6MB  | 13.87017       | 80.3     | 94.5     |
+| [DenseNet121](https://paddle-imagenet-models-name.bj.bcebos.com/DenseNet121_pretrained.tar) | 33.1MB   | 17.09874       | 75.7     | 92.6     |
+| [DenseNet161](https://paddle-imagenet-models-name.bj.bcebos.com/DenseNet161_pretrained.tar)| 118.0MB  | 22.79690       | 78.6     | 94.1     |
+| [DenseNet201](https://paddle-imagenet-models-name.bj.bcebos.com/DenseNet201_pretrained.tar)|  84.1MB   | 25.26089       | 77.6     | 93.7     |
+| [ShuffleNetV2](https://paddle-imagenet-models-name.bj.bcebos.com/ShuffleNetV2_pretrained.tar) | 10.2MB   | 15.40138        | 68.8     | 88.5     |
+| [HRNet_W18](https://paddle-imagenet-models-name.bj.bcebos.com/HRNet_W18_C_pretrained.tar) | 21.29MB |45.25514  | 76.9 | 93.4 |
 
 ## 目标检测模型
 
-> 表中模型相关指标均为在MSCOCO数据集上使用PaddlePaddle Python预测接口测试得到(测试GPU型号为Nvidia Tesla V100测试得到,表中符号`-`表示相关指标暂未测试。
+> 表中模型相关指标均为在MSCOCO数据集上使用PaddlePaddle Python预测接口测试得到(测试GPU型号为Nvidia Tesla V100测试得到,表中符号`-`表示相关指标暂未测试。
 
 | 模型    | 模型大小    | 预测时间(毫秒) | BoxAP(%) |
 |:-------|:-----------|:-------------|:----------|
-|FasterRCNN-ResNet50|135.6MB| 78.450 | 35.2 |
-|FasterRCNN-ResNet50_vd| 135.7MB | 79.523 | 36.4 |
-|FasterRCNN-ResNet101| 211.7MB | 107.342 | 38.3 |
-|FasterRCNN-ResNet50-FPN| 167.2MB | 44.897 | 37.2 |
-|FasterRCNN-ResNet50_vd-FPN|168.7MB | 45.773 | 38.9 |
-|FasterRCNN-ResNet101-FPN| 251.7MB | 55.782 | 38.7 |
-|FasterRCNN-ResNet101_vd-FPN |252MB | 58.785 | 40.5 |
-|FasterRCNN-HRNet_W18-FPN |115.5MB | 57.11 | 36 |
-|YOLOv3-DarkNet53|252.4MB | 21.944 | 38.9 |
-|YOLOv3-MobileNetv1 |101.2MB | 12.771 | 29.3 |
-|YOLOv3-MobileNetv3|94.6MB | - | 31.6 |
-| YOLOv3-ResNet34|169.7MB | 15.784 | 36.2 |
+|[FasterRCNN-ResNet50](https://paddlemodels.bj.bcebos.com/object_detection/faster_rcnn_r50_1x.tar)|136.0MB| 197.715 | 35.2 |
+|[FasterRCNN-ResNet50_vd](https://paddlemodels.bj.bcebos.com/object_detection/faster_rcnn_r50_vd_1x.tar)| 136.1MB | 475.700 | 36.4 |
+|[FasterRCNN-ResNet101](https://paddlemodels.bj.bcebos.com/object_detection/faster_rcnn_r101_1x.tar)| 212.5MB | 582.911 | 38.3 |
+|[FasterRCNN-ResNet50-FPN](https://paddlemodels.bj.bcebos.com/object_detection/faster_rcnn_r50_fpn_1x.tar)| 167.7MB | 83.189 | 37.2 |
+|[FasterRCNN-ResNet50_vd-FPN](https://paddlemodels.bj.bcebos.com/object_detection/faster_rcnn_r50_vd_fpn_2x.tar)|167.8MB | 128.277 | 38.9 |
+|[FasterRCNN-ResNet101-FPN](https://paddlemodels.bj.bcebos.com/object_detection/faster_rcnn_r101_fpn_1x.tar)| 244.2MB | 156.097 | 38.7 |
+|[FasterRCNN-ResNet101_vd-FPN](https://paddlemodels.bj.bcebos.com/object_detection/faster_rcnn_r101_vd_fpn_2x.tar) |244.3MB | 119.788 | 40.5 |
+|[FasterRCNN-HRNet_W18-FPN](https://paddlemodels.bj.bcebos.com/object_detection/faster_rcnn_hrnetv2p_w18_1x.tar) |115.5MB | 81.592 | 36 |
+|[YOLOv3-DarkNet53](https://paddlemodels.bj.bcebos.com/object_detection/yolov3_darknet.tar)|249.2MB | 42.672 | 38.9 |
+|[YOLOv3-MobileNetV1](https://paddlemodels.bj.bcebos.com/object_detection/yolov3_mobilenet_v1.tar) |99.2MB | 15.442 | 29.3 |
+|[YOLOv3-MobileNetV3_large](https://paddlemodels.bj.bcebos.com/object_detection/yolov3_mobilenet_v3.pdparams)|100.7MB | 143.322 | 31.6 |
+| [YOLOv3-ResNet34](https://paddlemodels.bj.bcebos.com/object_detection/yolov3_r34.tar)|170.3MB | 23.185 | 36.2 |
 
 ## 实例分割模型
 
 > 表中模型相关指标均为在MSCOCO数据集上测试得到。
+
+| 模型    | 模型大小    | 预测时间(毫秒) | mIoU(%) |
+|:-------|:-----------|:-------------|:----------|
+|DeepLabv3+-MobileNetV2_x1.0|-| - | - |
+|DeepLabv3+-Xception41|-| - | - |
+|DeepLabv3+-Xception65|-| - | - |
+|UNet|-| - | - |
+|HRNet_w18|-| - | - |