This is a demo code which contrasts the use of the conventional Regional Covariance Descriptors (RCovDs) against approximate infinite dimensional RCovDs obtained by the random Fourier features and the Nystrom method. The nearest neighbour and the kernel partial least squares classifiers are taken into account. The code is tested on Matlab R2013a.
To see the results, run the script RunMe.m. The Classification accuracies will be shown in your command window.
This file is provided without any warranty of fitness for any purpose. You can redistribute this file and/or modify it under the terms of the GNU General Public License (GPL) as published by the Free Software Foundation, either version 3 of the License or (at your option) any later version.
Please cite the following paper if your are using this code:
"Approximate Infinite-Dimensional Region Covariance Descriptors for Image Classification", Masoud Faraki, Mehrtash Harandi, and Fatih Porikli, 40th IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), Brisbane, Australia, April 19-24, 2015.
@inproceedings{faraki2015approximate, title={APPROXIMATE INFINITE-DIMENSIONAL REGION COVARIANCE DESCRIPTORS FOR IMAGE CLASSIFICATION}, author={Faraki, Masoud and Harandi, Mehrtash T and Porikli, Fatih}, booktitle={Acoustics, Speech and Signal Processing (ICASSP), 2015 IEEE International Conference on}, pages={1364--1368}, year={2015}, organization={IEEE} }