PDF-Extract-Kit/project/pdf2markdown/configs/pdf2markdown.yaml

inputs: assets/demo/formula_detection
outputs: outputs/pdf2markdown
visualize: True
merge2markdown: True
tasks:
  layout_detection:
    model: layout_detection_yolo
    model_config:
      img_size: 1024
      conf_thres: 0.25
      iou_thres: 0.45
      model_path: models/Layout/YOLO/doclayout_yolo_ft.pt
  formula_detection:
    model: formula_detection_yolo
    model_config:
      img_size: 1280
      conf_thres: 0.25
      iou_thres: 0.45
      batch_size: 1
      model_path: models/MFD/YOLO/yolo_v8_ft.pt
  formula_recognition:
    model: formula_recognition_unimernet
    model_config:
      batch_size: 128
      cfg_path: pdf_extract_kit/configs/unimernet.yaml
      model_path: models/MFR/unimernet_tiny
  ocr:
    model: ocr_ppocr
    model_config:
      lang: ch
      show_log: True
      det_model_dir: models/OCR/PaddleOCR/det/ch_PP-OCRv4_det
      rec_model_dir: models/OCR/PaddleOCR/rec/ch_PP-OCRv4_rec
      det_db_box_thresh: 0.3