Deep Convolutional Network Cascade for Facial Point Detection

Yi Sun 1,  Xiaogang Wang 2,3,  Xiaoou Tang 1,3
1Department of Information Engineering, The Chinese University of Hong Kong
2Department of Electronic Engineering, The Chinese University of Hong Kong
3Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences
{sy011, xtang}@ie.cuhk.edu.hk, xgwang@ee.cuhk.edu.hk


This web page provides the executable files and datasets of our CVPR 2013 paper [1], so that researchers can repeat our experiments or test our facial point detector on other datasets. The code and datasets are for research purposes only. If you use our code or datasets, please cite the paper [1]. The material provided on this web page is subject to change.


Code

Face detector: [Download]
The executable file for the face detector used in our paper [1] is provided. Please see the readme.txt in the downloaded package.

Point detector: [Download]
The executable file for the facial point detector used in our paper [1] is provided. Please see the readme.txt in the downloaded package.

Dataset

Training set: [Download]
It contains 5,590 LFW images and 7,876 other images downloaded from the web. The training set and validation set are defined in trainImageList.txt and testImageList.txt, respectively. Each line of these text files starts with the image name, followed by the boundary positions of the face bounding box retured by our face detector, then followed by the positions of the five facial points.

Testing set: [Download]
It contains the 1,521 BioID images, 781 LFPW training images, and 249 LFPW test images used in our testing, together with the text files recording the boundary positions of the face bounding box retured by our face detector for each dataset. A few images that our face detector failed are not listed in the text files. LFPW images are renamed for the convenience of processing.

Comparison with other methods

Comparison results: [Download]
The numerical results corresponding to Figure 6, 7 in Section 5.2 in our paper [1] are provided. Please see the readme.txt in the downloaded package.

Reference

[1] Y. Sun, X. Wang, and X. Tang. Deep Convolutional Network Cascade for Facial Point Detection. In Proceedings
   of IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2013. [PDF]

Last update: Mar. 2014.