---
dataset_info:
  features:
  - name: image
    dtype: image
  - name: bboxes
    sequence:
      sequence: float64
  - name: category_id
    sequence: int64
  - name: segmentation
    sequence:
      sequence:
        sequence: float64
  - name: area
    sequence: float64
  - name: pdf_cells
    list:
      list:
      - name: bbox
        sequence: float64
      - name: font
        struct:
        - name: color
          sequence: int64
        - name: name
          dtype: string
        - name: size
          dtype: float64
      - name: text
        dtype: string
  - name: metadata
    struct:
    - name: coco_height
      dtype: int64
    - name: coco_width
      dtype: int64
    - name: collection
      dtype: string
    - name: doc_category
      dtype: string
    - name: image_id
      dtype: int64
    - name: num_pages
      dtype: int64
    - name: original_filename
      dtype: string
    - name: original_height
      dtype: float64
    - name: original_width
      dtype: float64
    - name: page_hash
      dtype: string
    - name: page_no
      dtype: int64
  splits:
  - name: train
    num_bytes: 812161642.09009
    num_examples: 2000
  - name: test
    num_bytes: 199653537.00740147
    num_examples: 500
  - name: validation
    num_bytes: 288239466.2505779
    num_examples: 750
  download_size: 1257989448
  dataset_size: 1300054645.3480694
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
  - split: test
    path: data/test-*
  - split: validation
    path: data/validation-*
---