Local Sliced-Wasserstein Feature Sets for Illumination-invariant Face Recognition

This repository contains the Python implementation of Dicrete RCDT transform and its application on illumination-invariant face recognition (paper link: https://arxiv.org/abs/2202.10642). It is also a part of PyTransKit (link: https://github.com/rohdelab/PyTransKit).

DRCDT-NS based Illumination-invariant Face Recognition Demo

1. Download the The Extended Yale Face Database B from: http://vision.ucsd.edu/~leekc/ExtYaleDatabase/ExtYaleB.html (Please use the "Cropped Images")
2. Extract the folder to the local directory, the directory should have the structure as shown below:

Local directory
└───CroppedYale
|    └───yaleB01
│        │   yaleB02_P00A+000E+00.pgm
│        │   yaleB02_P00A+000E+20.pgm
│        │   ...
|    └───yaleB02
|    └───yaleB03
│    |...
│   NS_classifier_patch.py
│   rcdt_hog_feature_extraction.py
|   utility.py
|   drcdt_yale_notebook.ipynb
|   README.md

3. Run the Jupyter notebook: drcdt_yale_notebook.ipynb

Dependencies

Please see the "requirements.txt".

Citation

Please cite the following publication when publishing findings that benefit from the codes provided here.

@article{zhuang2022local,
title={Local Sliced-Wasserstein Feature Sets for Illumination-invariant Face Recognition},
author={Zhuang, Yan and Li, Shiying and Yin, Xuwang and Rubaiyat, Abu Hasnat Mohammad and Rohde, Gustavo K and others},
journal={arXiv preprint arXiv:2202.10642},
year={2022}
}