enum_class.py 1.7 KB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465
  1. class BlockType:
  2. IMAGE = 'image'
  3. TABLE = 'table'
  4. IMAGE_BODY = 'image_body'
  5. TABLE_BODY = 'table_body'
  6. IMAGE_CAPTION = 'image_caption'
  7. TABLE_CAPTION = 'table_caption'
  8. IMAGE_FOOTNOTE = 'image_footnote'
  9. TABLE_FOOTNOTE = 'table_footnote'
  10. TEXT = 'text'
  11. TITLE = 'title'
  12. INTERLINE_EQUATION = 'interline_equation'
  13. LIST = 'list'
  14. INDEX = 'index'
  15. DISCARDED = 'discarded'
  16. class ContentType:
  17. IMAGE = 'image'
  18. TABLE = 'table'
  19. TEXT = 'text'
  20. INTERLINE_EQUATION = 'interline_equation'
  21. INLINE_EQUATION = 'inline_equation'
  22. class CategoryId:
  23. Title = 0
  24. Text = 1
  25. Abandon = 2
  26. ImageBody = 3
  27. ImageCaption = 4
  28. TableBody = 5
  29. TableCaption = 6
  30. TableFootnote = 7
  31. InterlineEquation_Layout = 8
  32. InterlineEquationNumber_Layout = 9
  33. InlineEquation = 13
  34. InterlineEquation_YOLO = 14
  35. OcrText = 15
  36. LowScoreText = 16
  37. ImageFootnote = 101
  38. class MakeMode:
  39. MM_MD = 'mm_markdown'
  40. NLP_MD = 'nlp_markdown'
  41. CONTENT_LIST = 'content_list'
  42. class ModelPath:
  43. vlm_root_hf = "opendatalab/MinerU2.0-2505-0.9B"
  44. vlm_root_modelscope = "OpenDataLab/MinerU2.0-2505-0.9B"
  45. pipeline_root_modelscope = "OpenDataLab/PDF-Extract-Kit-1.0"
  46. pipeline_root_hf = "opendatalab/PDF-Extract-Kit-1.0"
  47. doclayout_yolo = "models/Layout/YOLO/doclayout_yolo_docstructbench_imgsz1280_2501.pt"
  48. yolo_v8_mfd = "models/MFD/YOLO/yolo_v8_ft.pt"
  49. unimernet_small = "models/MFR/unimernet_hf_small_2503"
  50. pytorch_paddle = "models/OCR/paddleocr_torch"
  51. layout_reader = "models/ReadingOrder/layout_reader"
  52. slanet_plus = "models/TabRec/SlanetPlus/slanet-plus.onnx"
  53. class SplitFlag:
  54. CROSS_PAGE = 'cross_page'
  55. LINES_DELETED = 'lines_deleted'