metadata
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-*