[ICASSP 2024] Part Representation Learning with Teacher-Student Decoder for Occluded Person Re-identification

The official repository for Part Representation Learning with Teacher-Student Decoder for Occluded Person Re-identification

News!

ICASSP 2024 Presentation Slides, Posters, and videos are available here

Meeting

Zoom link for poster section at Wed, 17 Apr, 13:10 - 15:10 (UTC +9)

Pipeline

Requirements

Installation

pip install -r requirements.txt
(we use /torch 1.7.0 /torchvision 0.7.0 /timm 0.3.2 /cuda 10.2 / 11G 2080ti or 24G 3090 for training and evaluation.
Note that we use torch.cuda.amp to accelerate speed of training which requires pytorch >=1.6)

Prepare Datasets

mkdir data

Download the person datasetsDukeMTMC-reID,Occluded-Duke, and Move the new dataset split file to the 'Occluded_Duke' folder. Download the human parsing labels provided by bpbreid on GDrive

mv new_gallery.txt data/Occluded_duke/
mv new_query.txt data/Occluded_duke/

data
├── dukemtmcreid
│   └── images ..
|   └── masks ..
├── Occluded_Duke
│   └── images ..
│   └── masks ..
|   └── new_gallery.txt
|   └── new_query.txt

Prepare ViT Pre-trained Models

You need to download the ImageNet pretrained transformer model : ViT-Base, ViT-Small, DeiT-Small, DeiT-Base

Training and Evaluation

We utilize 1 GPU for training.

sh run_tsd.sh

Acknowledgement

Codebase from TransReID , bpbreid

Citation

If you find this code useful for your research, please cite our paper

@inproceedings{gao2024part,
  title={Part Representation Learning with Teacher-Student Decoder for Occluded Person Re-identification},
  author={Gao, Shang; Yu, Chenyang; Zhang, Pingping and Lu, Huchuan},
  booktitle={ICASSP},
  year={2024}
} 

Contact

If you have any question, please feel free to contact us. E-mail: gs940601k@gmail.com