Demo code for CVPR 2016 paper: Learning a Discriminative Null Space for Person Re-identification
Download data from here and unzip it unzip data.zip.
It contains the LOMO feature [1] and kCCA feature [2] for VIPeR dataset.
run demo.m in Matlab.
We used the VIPeR data split provided by [2] in https://github.com/glisanti/KCCAReId.
For LOMO feature, we can get reported result 42.28% on VIPeR. (RBF kernel).
For kCCA feature, we can get 46.68% (CHI2 kernel), 45.92% (RBF kernel).
We can get reported score-level fusion result 51% on VIPeR.
Download the CMC curve on VIPeR, PRID, CUHK01, CUHK03 and Market1501 from here.
If you use this code in your research, please use the following BibTeX entry.
@inproceedings{zhang2016learning,
title={Learning a discriminative null space for person re-identification},
author={Zhang, Li and Xiang, Tao and Gong, Shaogang},
booktitle={Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition},
year={2016}
}
- [1] Person Re-identification by Local Maximal Occurrence Representation and Metric Learning. Shengcai Liao, Yang Hu, Xiangyu Zhu, and Stan Z. Li. IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2015.
- [2] Matching people across camera views using kernel canonical correlation analysis. Giuseppe Lisanti, Iacopo Masi, and Alberto Del Bimbo. International Conference on Distributed Smart Cameras (ICDSC), 2014.
- [3] Kernel Null Space Methods for Novelty Detection. Paul Bodesheim, Alexander Freytag, Erik Rodner, Michael Kemmler, and Joachim Denzler. IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2013.
