Multi-scale Semantic Correlation Mining network

Pytorch code for paper "Multi-scale Semantic Correlation Mining for Visible-Infrared Person Re-Identification(arxiv)

https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4761606".

1. Results

We adopt the CNN-based AGW [3] as backbone respectively.

*The results may exhibit fluctuations due to random splitting, and further improvement can be achieved by fine-tuning the hyperparameters.

2. Datasets

• RegDB [1]: The RegDB dataset can be downloaded from this website.

• SYSU-MM01 [2]: 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.

  python pre_process_sysu.py

• LLCM [5]: The LLCM dataset can be downloaded by sending a signed dataset release agreement copy to zhangyk@stu.xmu.edu.cn.

3. Training

Train MSCMNet by

python train.py --dataset sysu --gpu 0

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

• --gpu: which gpu to run.

You may need manually define the data path first.

4. Testing

Test a model on SYSU-MM01 dataset by

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

• --dataset: which dataset "sysu" or "regdb".
• --mode: "all" or "indoor" (only for sysu dataset).
• --resume: the saved model path.
• --gpu: which gpu to use.

Test a model on RegDB dataset by

python test.py --dataset 'regdb' --resume 'model_path'  --tvsearch True --gpu 0

• --tvsearch: whether thermal to visible search True or False (only for regdb dataset).

Test a model on LLCM dataset by

python test.py --dataset 'llcm' --resume 'model_path'  --gpu 0

5. Citation

Most of the code of our backbone are borrowed from AGW [3] and CAJ [4]. Most of the code related to LLCM dataset are borrowed from DEEN [5].

Thanks a lot for the author's contribution.

Please cite the following paper in your publications if it is helpful:

6. References.

[1] D. T. Nguyen, H. G. Hong, K. W. Kim, and K. R. Park. Person recognition system based on a combination of body images from visible light and thermal cameras. Sensors, 17(3):605, 2017.

[2] A. Wu, W.-s. Zheng, H.-X. Yu, S. Gong, and J. Lai. Rgb-infrared crossmodality person re-identification. In IEEE International Conference on Computer Vision (ICCV), pages 5380–5389, 2017.

[3] Ye M, Shen J, Lin G, et al. Deep learning for person re-identification: A survey and outlook[J]. IEEE transactions on pattern analysis and machine intelligence, 2021, 44(6): 2872-2893.

[4] Ye M, Ruan W, Du B, et al. Channel augmented joint learning for visible-infrared recognition[C]//Proceedings of the IEEE/CVF International Conference on Computer Vision. 2021: 13567-13576.

[5] Zhang Y, Wang H. Diverse Embedding Expansion Network and Low-Light Cross-Modality Benchmark for Visible-Infrared Person Re-identification[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2023: 2153-2162.