layout_parsing.yaml 2.6 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101
  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. use_formula_recognition: False
  7. SubModules:
  8. LayoutDetection:
  9. module_name: layout_detection
  10. model_name: RT-DETR-H_layout_17cls
  11. model_dir: null
  12. SubPipelines:
  13. DocPreprocessor:
  14. pipeline_name: doc_preprocessor
  15. use_doc_orientation_classify: True
  16. use_doc_unwarping: True
  17. SubModules:
  18. DocOrientationClassify:
  19. module_name: doc_text_orientation
  20. model_name: PP-LCNet_x1_0_doc_ori
  21. model_dir: null
  22. DocUnwarping:
  23. module_name: image_unwarping
  24. model_name: UVDoc
  25. model_dir: null
  26. GeneralOCR:
  27. pipeline_name: OCR
  28. text_type: general
  29. use_doc_preprocessor: False
  30. use_textline_orientation: False
  31. SubModules:
  32. TextDetection:
  33. module_name: text_detection
  34. model_name: PP-OCRv4_server_det
  35. model_dir: null
  36. limit_side_len: 960
  37. limit_type: max
  38. thresh: 0.3
  39. box_thresh: 0.6
  40. unclip_ratio: 2.0
  41. TextRecognition:
  42. module_name: text_recognition
  43. model_name: PP-OCRv4_server_rec
  44. model_dir: null
  45. batch_size: 6
  46. score_thresh: 0
  47. TableRecognition:
  48. pipeline_name: table_recognition
  49. use_layout_detection: False
  50. use_doc_preprocessor: False
  51. use_ocr_model: False
  52. SubModules:
  53. TableStructureRecognition:
  54. module_name: table_structure_recognition
  55. model_name: SLANet_plus
  56. model_dir: null
  57. SealRecognition:
  58. pipeline_name: seal_recognition
  59. use_layout_detection: False
  60. use_doc_preprocessor: False
  61. SubPipelines:
  62. SealOCR:
  63. pipeline_name: OCR
  64. text_type: seal
  65. use_doc_preprocessor: False
  66. use_textline_orientation: False
  67. SubModules:
  68. TextDetection:
  69. module_name: seal_text_detection
  70. model_name: PP-OCRv4_server_seal_det
  71. model_dir: null
  72. limit_side_len: 736
  73. limit_type: min
  74. thresh: 0.2
  75. box_thresh: 0.6
  76. unclip_ratio: 0.5
  77. TextRecognition:
  78. module_name: text_recognition
  79. model_name: PP-OCRv4_server_rec
  80. model_dir: null
  81. batch_size: 1
  82. score_thresh: 0
  83. FormulaRecognition:
  84. pipeline_name: formula_recognition
  85. use_layout_detection: False
  86. use_doc_preprocessor: False
  87. SubModules:
  88. FormulaRecognition:
  89. module_name: formula_recognition
  90. model_name: PP-FormulaNet-L
  91. model_dir: null
  92. batch_size: 5