metadata
dataset_info:
features:
- name: image
dtype: image
- name: id
dtype: int32
- name: width
dtype: int32
- name: height
dtype: int32
- name: annotations
dtype: string
splits:
- name: train
num_bytes: 20303582337.618
num_examples: 118287
- name: val
num_bytes: 832898047
num_examples: 5000
download_size: 20361586602
dataset_size: 21136480384.618
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: val
path: data/val-*