metadata
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
dataset_info:
features:
- name: image_url
dtype: string
- name: landmarks
struct:
- name: eye_left
sequence: float64
- name: eye_right
sequence: float64
- name: mouth_left
sequence: int64
- name: mouth_right
sequence: int64
- name: raw_name
dtype: string
- name: text
dtype: string
- name: aligned_name
dtype: string
- name: image
dtype: image
splits:
- name: train
num_bytes: 18209758516.333
num_examples: 21287
download_size: 17989395966
dataset_size: 18209758516.333