Pixel-wise Graph Attention Networks for Person Re-identification (PGANet)

Pixel-wise Graph Attention Networks for Person Re-identification. ACM MM2021, Poster.

The codes are expanded on [Strong baseline].

1 Pipeline

This article proposes a feature extraction module based on pixel-wise graph attention, which treats image pixels as nodes on the graph and pixels surrounding them as neighbors. The graph attention mechanism is used to update image features. We insert this module into each layer in ResNet to enhance the components extracted by ResNet. The module details are as follows:

The following figure is the architecture diagram of the model:

2 Get Started

2.1 Environment

PyTorch, pytorch_ignite==0.1.2, yacs, and soon.

2.2 Dataset

Create a directory to store reid datasets under this repo or outside this repo. Remember to set your path to the root of the dataset in config/defaults.py for all training and testing or set it in every config file in configs/ or in every single command.

You can create a directory to store reid datasets under this repo via

（1）Market1501

Download dataset to data/ from http://www.liangzheng.org/Project/project_reid.html
Extract dataset and rename to market1501. The data structure would like:

data
    market1501 # this folder contains 6 files.
        bounding_box_test/
        bounding_box_train/
        ......

（2）DukeMTMC-reID

Download dataset to data/ from https://github.com/layumi/DukeMTMC-reID_evaluation#download-dataset
Extract dataset and rename to dukemtmc-reid. The data structure would like:

data
    dukemtmc-reid
        DukeMTMC-reID # this folder contains 8 files.
            bounding_box_test/
            bounding_box_train/
            ......

2.3 Prepare pretrained model

Download the corresponding model parameters from the series website, save them in the desired location, and change the location of the pre-trained model '_C.MODEL.PRETRAIN_PATH' in '/config/defaults.py' and 'configs/*.yml'.

model_urls = {
    'resnet50': 'https://download.pytorch.org/models/resnet50-19c8e357.pth',
    'resnet101': 'https://download.pytorch.org/models/resnet101-5d3b4d8f.pth',
    'resnet152': 'https://download.pytorch.org/models/resnet152-b121ed2d.pth',
}

3 Train

You can run these commands in .sh files for training different datasets of different loss. You can also directly run code sh *.sh to run our demo after your custom modification.

Experiment-all_tricks-tri_center-market-pga.sh：

python3 tools/train.py --config_file='configs/softmax_triplet_with_center_pga.yml' SOLVER.MAX_EPOCHS "(600)"  MODEL.DEVICE_ID "('1')" DATASETS.NAMES "('market1501')" DATASETS.ROOT_DIR "('your location')" MODEL.SGA_ON "('True')" MODEL.CGA_ON "('False')" MODEL.NUM_HEADS "(1)" OUTPUT_DIR "('your output log location')"

4 Test

Test-pga-all_tricks-tri_center-market.sh：

python3 tools/test.py --config_file='configs/softmax_triplet_with_center_pga.yml' MODEL.DEVICE_ID "('0')" DATASETS.NAMES "('market1501')" DATASETS.ROOT_DIR "('your location')" MODEL.PRETRAIN_CHOICE "('self')" TEST.WEIGHT "('your logcation/model.pth')"

5 Related Works

· Bag of Tricks and A Strong Baseline for Deep Person Re-identification [Paper] [Code]

wenyu1009 / PGANet