enum_class.py 1.7 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566
  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. EQUATION = 'equation'
  23. class CategoryId:
  24. Title = 0
  25. Text = 1
  26. Abandon = 2
  27. ImageBody = 3
  28. ImageCaption = 4
  29. TableBody = 5
  30. TableCaption = 6
  31. TableFootnote = 7
  32. InterlineEquation_Layout = 8
  33. InterlineEquationNumber_Layout = 9
  34. InlineEquation = 13
  35. InterlineEquation_YOLO = 14
  36. OcrText = 15
  37. LowScoreText = 16
  38. ImageFootnote = 101
  39. class MakeMode:
  40. MM_MD = 'mm_markdown'
  41. NLP_MD = 'nlp_markdown'
  42. CONTENT_LIST = 'content_list'
  43. class ModelPath:
  44. vlm_root_hf = "opendatalab/MinerU2.0-2505-0.9B"
  45. vlm_root_modelscope = "OpenDataLab/MinerU2.0-2505-0.9B"
  46. pipeline_root_modelscope = "OpenDataLab/PDF-Extract-Kit-1.0"
  47. pipeline_root_hf = "opendatalab/PDF-Extract-Kit-1.0"
  48. doclayout_yolo = "models/Layout/YOLO/doclayout_yolo_docstructbench_imgsz1280_2501.pt"
  49. yolo_v8_mfd = "models/MFD/YOLO/yolo_v8_ft.pt"
  50. unimernet_small = "models/MFR/unimernet_hf_small_2503"
  51. pytorch_paddle = "models/OCR/paddleocr_torch"
  52. layout_reader = "models/ReadingOrder/layout_reader"
  53. slanet_plus = "models/TabRec/SlanetPlus/slanet-plus.onnx"
  54. class SplitFlag:
  55. CROSS_PAGE = 'cross_page'
  56. LINES_DELETED = 'lines_deleted'