---
dataset_info:
  features:
  - name: image
    dtype: image
  - name: caption
    dtype: string
  - name: annotations
    struct:
    - name: polygon
      sequence:
        sequence:
          sequence: int64
    - name: text
      sequence: string
    - name: valid
      sequence: bool
    - name: language
      sequence: string
    - name: illegibility
      sequence: bool
    - name: rec_score
      sequence: float32
    - name: pos
      sequence: int64
  splits:
  - name: train
    num_bytes: 456723076.25
    num_examples: 6654
  download_size: 454285438
  dataset_size: 456723076.25
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
---