Support Neighbor Loss for Person Re-Identification

This repository is for the paper introduced in the following paper

Kai Li, Zhengming Ding, Kunpeng Li, Yulun Zhang, and Yun Fu, "Support Neighbor Loss for Person Re-Identification", ACM Multimedia (ACM MM) 2018, [arXiv]

Environment

Python 3 + PyTorch 3.0

Data preparation

Please refer this repo for the data preparation and modify the data locations accordingly in the train.sh and test.sh files.

Train

sh ./train.sh

Test

Pretrained models: Market1501, CUHK01, CUHK03

sh ./test.sh

Results

Quantitative Results

Visual Results

Citation

If you find the code helpful in your resarch or work, please cite the following papers.

@inproceedings{li2018support,
  title={Support neighbor loss for person re-identification},
  author={Li, Kai and Ding, Zhengming and Li, Kunpeng and Zhang, Yulun and Fu, Yun},
  booktitle={2018 ACM Multimedia Conference on Multimedia Conference},
  pages={1492--1500},
  year={2018},
  organization={ACM}
}

Acknowledgements

This code is built on this repository, developed by Houjing Huang.