Skip to content
Branch: master
Find file History
Latest commit e32b421 Aug 13, 2019
Permalink
Type Name Latest commit message Commit time
..
Failed to load latest commit information.
models update TS3 computation cost Aug 13, 2019
README.md update TS3 computation cost Aug 13, 2019
test.py update TS3 computation cost Aug 13, 2019

README.md

Teacher Supervises Students How to Learn from Partially Labeled Images for Facial Landmark Detection

By Xuanyi Dong, Yi Yang

This paper presents a semi-supervised facial landmark detection algorithm. In short, we propose to learn a teacher network that takes a pseudo labeled face as input and outputs the quality of its pseudo label. As a result, pseudo labeled samples which are qualified by the teacher will be used to train the student detector. We achieve higher accuracy than typical semi-supervised facial landmark methods.

Model Configs

We released the model definition files in the models folder, including two student detectors and one teacher network.

Please use the following script to check the model size and FLOPs of each student detector.

python test.py

CPM : 16.70 MB parameters and 1720.98 M FLOPs.

HG : 24.97 MB parameters and 1600.85 M FLOPs.

Full Project Codes?

This project was done at about June 2018. The original codes use an old PyTorch version (0.3.1) and are a little bit messy. I'm currently re-organizing codes and moving it to PyTorch 1.0.1. Thanks for your patience for waiting.

Citation

If this project helps your research, please cite the following papers:

@inproceedings{dong2019teacher,
  title={Teacher Supervises Students How to Learn from Partially Labeled Images for Facial Landmark Detection},
  author={Dong, Xuanyi and Yang, Yi},
  booktitle={Proceedings of the IEEE International Conference on Computer Vision (ICCV)},
  year={2019}
}

Contact

To ask questions or report issues, please open an issue on the issues tracker.

You can’t perform that action at this time.