[TIP2020] Adaptive Graph Representation Learning for Video Person Re-identification

Introduction

This is the PyTorch implementation for Adaptive Graph Representation Learning for Video Person Re-identification.

Get started

git clone https://github.com/weleen/AGRL.pytorch /path/to/save
pip install -r requirements.txt
cd torchreid/metrics/rank_cylib && make

Dataset

create dataset directory

mkdir data

Prepare datasets:

├── dukemtmc-vidreid
│   ├── DukeMTMC-VideoReID
│   ├── pose.json
│   ├── split_gallery.json
│   ├── split_query.json
│   └── split_train.json
│
├── ilids-vid
│   ├── i-LIDS-VID
│   ├── pose.json
│   ├── splits.json
│   └── train-test people splits
│
├── mars
│   ├── bbox_test
│   ├── bbox_train
│   ├── info
│   ├── pose.json
│   └── train-test people splits
│
├── prid2011
    ├── pose.json
    ├── prid_2011
    ├── prid_2011.zip
    ├── splits_prid2011.json
    └── train_test_splits_prid.mat

pose.json is obtained by running AlphaPose, we put the files on Baidu Netdisk (code: luxr) and Google Driver.

More details could be found in DATASETS.md.

Train

bash scripts/train_vidreid_xent_htri_vmgn_mars.sh

To use multiple GPUs, you can set --gpu-devices 0,1,2,3.

Note: To resume training, you can use --resume path/to/model to load a checkpoint from which saved model weights and start_epoch will be used. Learning rate needs to be initialized carefully. If you just wanna load a pretrained model by discarding layers that do not match in size (e.g. classification layer), use --load-weights path/to/model instead.

Please refer to the code for more details.

Test

create a directory to store model weights mkdir saved-models/ beforehand. Then, run the following command to test

bash scripts/test_vidreid_xent_htri_vmgn_mars.sh

All the model weights are available.

Model

All the results tested with 4 TITAN X GPU and 64GB memory.

Citation

Please kindly cite this project in your paper if it is helpful 😊 :

@article{wu2020adaptive,
  title={Adaptive graph representation learning for video person re-identification},
  author={Wu, Yiming and Bourahla, Omar El Farouk and Li, Xi* and Wu, Fei and Tian, Qi and Zhou, Xue},
  journal={IEEE Transactions on Image Processing},
  year={2020},
  publisher={IEEE}
}

This project is developed based on deep-person-reid and STE-NVAN.