metadata
dataset_info:
features:
- name: image
dtype: image
- name: image_id
dtype: string
- name: objects
struct:
- name: category_id
sequence: int64
- name: bbox
sequence:
sequence: float64
- name: text_input
sequence: string
splits:
- name: train
num_bytes: 2593865
num_examples: 40
download_size: 2596516
dataset_size: 2593865
configs:
- config_name: default
data_files:
- split: train
path: data/train-*