CPR

Official PyTorch implementation of CPR

Datasets

We use the MVTec AD dataset for experiments. And use DTD data set to simulate anomalous image.

The data directory is as follows:

data
├── dtd
│   ├── images
│   ├── imdb
│   └── labels
└── mvtec
    ├── bottle
    │   ├── ground_truth
    │   ├── license.txt
    │   ├── readme.txt
    │   ├── test
    │   └── train
    ...
    └── zipper
        ├── ground_truth
        ├── license.txt
        ├── readme.txt
        ├── test
        └── train

Installation

pytorch

conda install pytorch torchvision torchaudio pytorch-cuda=11.8 -c pytorch -c nvidia

pip install -r requirements.txt

generate foreground and global retrieval result

python tools/generate_foreground.py
python tools/generate_retrieval.py

Training

generate synthetic data

python tools/generate_synthetic_data.py -fd log/foreground_mvtec_DenseNet_features.denseblock1_320

bash train.sh

Testing

python test.py -fd log/foreground_mvtec_DenseNet_features.denseblock1_320/ --checkpoints weights/{category}.pth

Pretrained Checkpoints

Download pretrained checkpoints here and put the checkpoints under <project_dir>/weights/.

Baidu Netdisk: https://pan.baidu.com/s/1FTE4b2G8nVZt4lUyaP-kIQ?pwd=ky7j

Acknowledgement

We borrow some codes from PatchCore, MemSeg and SuperPoint

Citation

@ARTICLE{10678861,
  author={Li, Hanxi and Hu, Jianfei and Li, Bo and Chen, Hao and Zheng, Yongbin and Shen, Chunhua},
  journal={IEEE Transactions on Image Processing}, 
  title={Target Before Shooting: Accurate Anomaly Detection and Localization Under One Millisecond via Cascade Patch Retrieval}, 
  year={2024},
  volume={33},
  number={},
  pages={5606-5621},
  keywords={Accuracy;Anomaly detection;Measurement;Standards;Prototypes;Image retrieval;Image reconstruction;Anomaly detection;image patch retrieval;metric learning},
  doi={10.1109/TIP.2024.3448263}
}