metadata
dataset_info:
features:
- name: id
dtype: float64
- name: brand
dtype: string
- name: image_id
dtype: string
- name: caption
dtype: string
- name: image_url
dtype: string
- name: width
dtype: string
- name: height
dtype: string
- name: image
dtype: image
- name: num_people
dtype: int64
- name: mask_metadata
list:
- name: box
sequence: float64
- name: label
dtype: string
- name: label_id
dtype: int64
- name: polygons
sequence:
sequence: int64
- name: score
dtype: float64
splits:
- name: train
num_bytes: 9940110798.883486
num_examples: 23279
download_size: 8810290815
dataset_size: 9940110798.883486
configs:
- config_name: default
data_files:
- split: train
path: data/train-*