vehicle_reid_by_parsing

This repo gives the code for the paper "Xinchen Liu, Wu Liu, Jinkai Zheng, Chenggang Yan, Tao Mei: Beyond the Parts: Learning Multi-view Cross-part Correlation for Vehicle Re-identification. ACM MM 2020". This code is based on reid strong baseline.

Requirements

• Linux or macOS with python ≥ 3.6
• PyTorch ≥ 1.0
• torchvision that matches the Pytorch installation. You can install them together at pytorch.org to make sure of this.
• yacs
• Cython (optional to compile evaluation code)
• tensorboard (needed for visualization): pip install tensorboard

Data Preparation

To train a vehicle reid model with parsing, you need the original image datasets like VeRi and the parsing masks of all images. For vehicle parsing models pretrained on the MVP dataset based on PSPNet/DeepLabV3/HRNet, please refer to this repo.

Training

You can run the examplar training script in .sh files.

Main Code

The main code for GCN can be found in

root
  engine
    trainer_selfgcn.py    # training pipline
  modeling
    baseline_selfgcn.py   # definition of the model
  tools
    train_selfgcn.py      # training preparation

The code for data io and sampler also be modified for the parsing based reid method.

License

PCRNet is released under the Apache 2.0 license.

Reference

@inproceedings{mm/LiuLZY020,
  author    = {Xinchen Liu and
               Wu Liu and
               Jinkai Zheng and
               Chenggang Yan and
               Tao Mei},
  title     = {Beyond the Parts: Learning Multi-view Cross-part Correlation for Vehicle
               Re-identification},
  booktitle = {ACM MM},
  pages     = {907--915},
  year      = {2020}
}