A Fourier-based Semantic Augmentation for Visible-Thermal Person Re-Identification

Pytorch Code of FSA method for Cross-Modality Person Re-Identification (Visible Thermal Re-ID) on RegDB dataset and SYSU-MM01 dataset.

Results

Dataset	Rank1	mAP
RegDB	~87.23%	~80.70%
SYSU-MM01	~73.63%	~69.25%

Usage

Our code extends the pytorch implementation of Cross-Modal-Re-ID-baseline in Github. Please refer to the offical repo for details of data preparation.

Training

Train a model by

python train_ext.py --dataset sysu --lr 0.1 --batch-size 6 --num_pos 4 --fsa_method FSA --lam 0.8 --gpu 0

• --dataset: which dataset "sysu" or "regdb".

• --lr: initial learning rate.

• --gpu: which gpu to run.

• --fsa_method: which semantic augmentation method to use.

Testing

Test a model on SYSU-MM01 or RegDB dataset by using testing augmentation with HorizontalFlip

python testa.py --mode all --resume 'model_path' --gpu 0 --dataset sysu

• --dataset: which dataset "sysu" or "regdb".

• --mode: "all" or "indoor" all search or indoor search (only for sysu dataset).

• --trial: testing trial (only for RegDB dataset).

• --resume: the saved model path.

• --gpu: which gpu to run.

Citation

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

@article{liu2020parameter,
  title={Parameter sharing exploration and hetero-center triplet loss for visible-thermal person re-identification},
  author={Liu, Haijun and Tan, Xiaoheng and Zhou, Xichuan},
  journal={IEEE Transactions on Multimedia},
  volume={23},
  pages={4414--4425},
  year={2020},
  publisher={IEEE}
}

@article{liu2021strong,
  title={Strong but simple baseline with dual-granularity triplet loss for visible-thermal person re-identification},
  author={Liu, Haijun and Chai, Yanxia and Tan, Xiaoheng and Li, Dong and Zhou, Xichuan},
  journal={IEEE Signal Processing Letters},
  volume={28},
  pages={653--657},
  year={2021},
  publisher={IEEE}
}

@article{Tan2022AFS,
  title={A Fourier-Based Semantic Augmentation for Visible-Thermal Person Re-Identification},
  author={Xiaoheng Tan and Yanxia Chai and Fenglei Chen and Haijun Liu},
  journal={IEEE Signal Processing Letters},
  year={2022},
  volume={29},
  pages={1684-1688}
}