In this project, we will release our deep person re-identification (Re-ID) implementation based on caffe. In the examples file, you can find the corresponding algorithms, which are introduced as follows:
Three main contributions are made in this paper: 1) A part-based deep neural network is designed for person Re-ID; 2) A point to set metric is introduced to supervise feature learning; 3) A symmetric triplet loss function is formulated to jointly minimize the intra-class distance and maximize the inter-class distance.
Please cite our paper if it helps your research:
@inproceedings{zhou2017point,
title={Point to set similarity based deep feature learning for person re-identification},
author={Zhou, Sanping and Wang, Jinjun and Wang, Jiayun and Gong, Yihong and Zheng, Nanning},
booktitle={Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition},
pages={3741--3750},
year={2017}
}