metadata
dataset_info:
features:
- name: file_name
dtype: string
- name: caption
dtype: string
- name: bbox
dtype: string
- name: category
dtype: string
splits:
- name: train
num_bytes: 2646007
num_examples: 22407
download_size: 637710
dataset_size: 2646007
configs:
- config_name: default
data_files:
- split: train
path: data/train-*