Skip to content
Multi-Objective Convolutional Network for Face Labeling
C++ Python Cuda CMake Protocol Buffer Makefile Other
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
cmake
data/LFW
docs
examples/LFW_cvpr15
include/caffe
matlab
python
scripts
src
tools
.Doxyfile
.gitattributes
.gitignore
.travis.yml
CMakeLists.txt
CONTRIBUTING.md
CONTRIBUTORS.md
INSTALL.md
LICENSE
Makefile
Makefile.config
README.md
caffe.cloc

README.md

Please check https://github.com/Liusifei/Face_Parsing_2016 for latest HELEN (11 classes) model.

Since we change the matlab interface, you need to adapt the Makefile.config to your own machine.

Compile caffe and matlab respectively using "make all -j8" and "make matcaffe" at FaceLabeling/;

Run Face3_demo.m (cd FaceLabeling/matlab/caffe/lfw_cvpr15) to obtain the 3-classes parsing result.

If you want to test other facial images, please include a shape file (e.g. shape.txt) that contains the face's five points positions. There are many existing tools to generate the 5pts, e.g. http://mmlab.ie.cuhk.edu.hk/archive/CNN_FacePoint.htm

The Helen model will be released soon.

Please cite this work in your publications if it helps your research:

@inproceedings{liu2015multi,
  Author = {Sifei Liu and Jimei Yang and Chang Huang and Ming-Hsuan Yang},
  booktitle={Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition},
  Title = {Multi-Objective Convolutional Learning for Face Labeling},
  Year = {2015}
}
You can’t perform that action at this time.