Spectral Aware Softmax for Visible-Infrared Person Re-Identification

The official repository for Spectral Aware Softmax for Visible-Infrared Person Re-Identification [pdf]

Prepare Datasets

(1) RegDB Dataset [3]: The RegDB dataset can be downloaded from this website by submitting a copyright form.
- (Named: "Dongguk Body-based Person Recognition Database (DBPerson-Recog-DB1)" on their website).
(2) SYSU-MM01 Dataset [4]: The SYSU-MM01 dataset can be downloaded from this website.
- run python pre_process_sysu.py to pepare the dataset, the training data will be stored in ".npy" format.

Training

We utilize 1 3090 GPU for training and you can train the SA-Softmax with:

python train_sas.py --gpu 'your device id' --dataset 'sysu or regdb'

Some examples:

# SYSU-MM01
python train_sas.py --gpu 0 --dataset sysu

Evaluation

python test.py --mode 'mode for SYSU-MM01' --resume 'model_path' --gpu 'your device id' --dataset 'sysu or regdb'

Some examples:

# SYSU-MM01
python test.py --mode all --resume SAS_sysu.t --gpu 0 --dataset sysu

Citation

Please kindly cite this paper in your publications if it helps your research:

@article{tan2023spectral,
  title={Spectral Aware Softmax for Visible-Infrared Person Re-Identification},
  author={Tan, Lei and Dai, Pingyang and Ye, Qixiang and Xu, Mingliang and Wu, Yongjian and Ji, Rongrong},
  journal={arXiv preprint arXiv:2302.01512},
  year={2023}
}

Acknowledgement

Our code is based on Cross-Modal-Re-ID-baseline[1, 2]

References

[1] M. Ye, J. Shen, G. Lin, T. Xiang, L. Shao, and S. C., Hoi. Deep learning for person re-identification: A survey and outlook. IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI), 2020.

[2] M. Ye, X. Lan, Z. Wang, and P. C. Yuen. Bi-directional Center-Constrained Top-Ranking for Visible Thermal Person Re-Identification. IEEE Transactions on Information Forensics and Security (TIFS), 2019.

Contact

If you have any question, please feel free to contact us. E-mail: tanlei@stu.xmu.edu.cn

stone96123/SA-Softmax