metadata
dataset_info:
features:
- name: image
dtype: image
- name: original_index
dtype: int64
- name: landmark
sequence:
sequence: int64
- name: mask
dtype: 'null'
splits:
- name: train
num_bytes: 2042785562
num_examples: 26904
download_size: 2004161055
dataset_size: 2042785562
configs:
- config_name: default
data_files:
- split: train
path: data/train-*