Skip to content
rectified version of the original HELEN dataset.
Branch: master
Clone or download
Jinpeng Lin (FA Talent)
Jinpeng Lin (FA Talent) update url
Latest commit c90b754 Jul 10, 2019
Type Name Latest commit message Commit time
Failed to load latest commit information. update url Jul 10, 2019

The HELEN* Dataset

The HELEN* dataset is a rectified version of the original HELEN dataset.

You could download the dataset using ( Google Drive | Baidu Drive | Dropbox )

File structure:

train/ # the training set
    xxx_image.jpg # the face image
    xxx_label.png # the mask annotations
    xxx_viz.jpg   # annotation visualization
test/ # the testing set (in original size)
test_resize/ # the testing set (resized)

All images in train/ come from the original HELEN. We relabeled the hair and facial skin regions in train/. Annotations in test/ also come from the original HELEN. While annotations in test_resize/ comply with the resized HELEN.

Mask annotation can be loaded by

import cv2
labels = cv2.imread('xxx_label.png', cv2.IMREAD_GRAYSCALE)

Meaning of annotation values:

0: background
1: facial skin
2: left brow (viewer side)
3: right brow
4: left eye
5: right eye
6: nose
7: upper lip
8: inner mouth
9: lower lip
10: hair

F scores can be calculated using


If you find the dataset useful, please consider citing

  title={Face Parsing with RoI Tanh-Warping},
  author={Lin, Jinpeng and Yang, Hao and Chen, Dong and Zeng, Ming and Wen, Fang and Yuan, Lu},
  booktitle={Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition},
You can’t perform that action at this time.