Intro

This repository provides MATLAB codes for patch learning and patch plots as shown in our previous publications:

[1] "Joint Patch and Multi-label Learning for Facial Action Detection", CVPR 2016.

[2] "Joint Patch and Multi-label Learning for Facial Action Unit and Holistic Expression Recognition", TIP 2016.

This code aims for active patch learning (APL), and the learned patches display in face images. This code requires "SPArse Modeling Software".

File structure

APL:

• Patch learning based on SPAMS toolbox.
• Show the patch importance in mean face shape.

More info

• Contact: Please send comments to Kaili Zhao (kailizhao@bupt.edu.cn)
• Citation: If you use this code in your paper, please cite one of the following:

@ARTICLE{zhao2016joint, 
  author = {Zhao, Kaili and Chu, Wen-Sheng and {De la Torre}, Fernando and Cohn, Jeffrey F. and Zhang, Honggang},
  journal={IEEE Transactions on Image Processing}, 
  title={Joint Patch and Multi-label Learning for Facial Action Unit and Holistic Expression Recognition}, 
  year={2016}, 
  volume={25}, 
  number={8}, 
  pages={3931-3946}
}

@InProceedings{zhao2015joint,
  author = {Zhao, Kaili and Chu, Wen-Sheng and {De la Torre}, Fernando and Cohn, Jeffrey F. and Zhang, Honggang},
  title = {Joint Patch and Multi-Label Learning for Facial Action Unit Detection},
  booktitle = {CVPR},
  year = {2015}
}