layout_parsing.yaml 2.3 KB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768697071727374757677787980818283848586878889
  1. pipeline_name: layout_parsing
  2. use_doc_preprocessor: True
  3. use_general_ocr: True
  4. use_seal_recognition: True
  5. use_table_recognition: True
  6. SubModules:
  7. LayoutDetection:
  8. module_name: layout_detection
  9. model_name: RT-DETR-H_layout_3cls
  10. model_dir: null
  11. SubPipelines:
  12. DocPreprocessor:
  13. pipeline_name: doc_preprocessor
  14. use_doc_orientation_classify: True
  15. use_doc_unwarping: True
  16. SubModules:
  17. DocOrientationClassify:
  18. module_name: doc_text_orientation
  19. model_name: PP-LCNet_x1_0_doc_ori
  20. model_dir: null
  21. DocUnwarping:
  22. module_name: image_unwarping
  23. model_name: UVDoc
  24. model_dir: null
  25. GeneralOCR:
  26. pipeline_name: OCR
  27. text_type: general
  28. use_doc_preprocessor: False
  29. use_textline_orientation: False
  30. SubModules:
  31. TextDetection:
  32. module_name: text_detection
  33. model_name: PP-OCRv4_server_det
  34. model_dir: null
  35. limit_side_len: 960
  36. limit_type: max
  37. thresh: 0.3
  38. box_thresh: 0.6
  39. unclip_ratio: 2.0
  40. TextRecognition:
  41. module_name: text_recognition
  42. model_name: PP-OCRv4_server_rec
  43. model_dir: null
  44. batch_size: 1
  45. score_thresh: 0
  46. TableRecognition:
  47. pipeline_name: table_recognition
  48. use_layout_detection: False
  49. use_doc_preprocessor: False
  50. use_ocr_model: False
  51. SubModules:
  52. TableStructureRecognition:
  53. module_name: table_structure_recognition
  54. model_name: SLANet_plus
  55. model_dir: null
  56. SealRecognition:
  57. pipeline_name: seal_recognition
  58. use_layout_detection: False
  59. use_doc_preprocessor: False
  60. SubPipelines:
  61. SealOCR:
  62. pipeline_name: OCR
  63. text_type: seal
  64. use_doc_preprocessor: False
  65. use_textline_orientation: False
  66. SubModules:
  67. TextDetection:
  68. module_name: seal_text_detection
  69. model_name: PP-OCRv4_server_seal_det
  70. model_dir: null
  71. limit_side_len: 736
  72. limit_type: min
  73. thresh: 0.2
  74. box_thresh: 0.6
  75. unclip_ratio: 0.5
  76. TextRecognition:
  77. module_name: text_recognition
  78. model_name: PP-OCRv4_server_rec
  79. model_dir: null
  80. batch_size: 1
  81. score_thresh: 0