Home

Awesome

DDAG

Pytorch Code of DDAG for Visible-Infrared Person Re-Identification in ECCV 2020. PDF

A Huawei MindSpore implementation of our DDAG method is HERE. Thanks to Zhiwei Zhang zhangzw12319@163.com.

Highlight

The goal of this work is to learn a robust and discriminative cross-modality representation for visible-infrarerd person re-identification.

Results on the SYSU-MM01 Dataset

MethodDatasetsRank@1mAPmINP
AGW [1]#SYSU-MM01 (All-Search)~ 47.50%~ 47.65%~ 35.30%
DDAG#SYSU-MM01 (All-Search)~ 54.75%~ 53.02%~39.62%
AGW [1]#SYSU-MM01 (Indoor-Search)~ 54.17%~ 62.97%~ 59.23%
DDAG#SYSU-MM01 (Indoor-Search)~ 61.02%~ 67.98%~ 62.61%

*The code has been tested in Python 3.7, PyTorch=1.0. Both of these two datasets may have some fluctuation due to random spliting

1. Prepare the datasets.

2. Training.

Train a model by

python train_ddag.py --dataset sysu --lr 0.1 --graph --wpa --part 3 --gpu 0

You may need manually define the data path first.

3. Testing.

Test a model on SYSU-MM01 or RegDB dataset by

python test_ddag.py --dataset sysu --mode all --wpa --graph --gpu 1 --resume 'model_path' 

4. Citation

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

@inproceedings{eccv20ddag,
  title={Dynamic Dual-Attentive Aggregation Learning for Visible-Infrared Person Re-Identification},
  author={Ye, Mang and Shen, Jianbing and Crandall, David J. and Shao, Ling and Luo, Jiebo},
  booktitle={European Conference on Computer Vision (ECCV)},
  year={2020},
}
@article{arxiv20reidsurvey,
  title={Deep Learning for Person Re-identification: A Survey and Outlook},
  author={Ye, Mang and Shen, Jianbing and Lin, Gaojie and Xiang, Tao and Shao, Ling and Hoi, Steven C. H.},
  journal={arXiv preprint arXiv:2001.04193},
  year={2020},
}

Contact: mangye16@gmail.com