Something You should know

The official code for SCAN is here.

The Optical Flow datasets for ilids and prid2011 are not avaliable, you should generate it by yourself,or you can choose don't use the Optical flow.

You can get the optical_flow data for ilids_vid from here https://drive.google.com/open?id=1u9jMd9wmmW25fAKGTRWg6pPRJtemcp4v, which is provided by https://github.com/dapengchen123/video_reid.

Don't forget change the optical_flow data name to make sure it's same as in ilidsvidsequence.py or prid2011sequence.py.

What I do

1.fix some bugs to make code runable on

pytorch>=0.4, Python>=3.6

2.the files that I have changed some where:

    a. seqtransforms.py
    b. sampler.py
    c. attevaluator.py
    d. classifer.py

Self-and-Collaborative Attention Network

This solution contains source code of the project "SCAN: Self-and-Collaborative Attention Network for Video Person Re-identification"

The source code is for educational and research use only without any warranty; if you use any part of the source code, please cite related paper:

@article{zhang2018scan,
  title={SCAN: Self-and-Collaborative Attention Network for Video Person Re-identification},
  author={Zhang, Ruimao and Sun, Hongbin and Li, Jingyu and Ge, Yuying and Lin, Liang and Luo, Ping and Wang, Xiaogang},
  journal={IEEE Trans. on Image Processing},
  year={2019}
}

License

All materials in this repository are released under the CC-BY-NC 4.0 LICENSE.