metadata
dataset_info:
features:
- name: image
dtype: image
- name: caption
dtype: string
- name: annotations
sequence:
- name: polygon
sequence:
sequence: int32
- name: text
dtype: string
- name: valid
dtype: int32
- name: language
dtype: string
- name: illegibility
dtype: int32
- name: rec_score
dtype: float32
- name: pos
dtype: int32
splits:
- name: train
num_bytes: 8217826
num_examples: 176
download_size: 8095811
dataset_size: 8217826
configs:
- config_name: default
data_files:
- split: train
path: data/train-*