PBAS

Progressive Boundary Guided Anomaly Synthesis for Industrial Anomaly Detection

Qiyu Chen, Huiyuan Luo, Han Gao, Chengkan Lv*, Zhengtao Zhang

IEEE DOI Link & ArXiv Preprint Link

Table of Contents

• 📖 Introduction
• 🔧 Environments
• 📊 Data Preparation
• 🚀 Run Experiments
• 🔗 Citation
• 🙏 Acknowledgements
• 📜 License

Introduction

This repository provides PyTorch-based source code for PBAS, a framework that enhances unsupervised anomaly detection by directionally synthesizing significant anomalies without predefined texture properties, guided by a progressive decision boundary.

Environments

Create a new conda environment and install required packages.

conda create -n pbas_env python=3.9.15
conda activate pbas_env
pip install -r requirements.txt

Experiments were conducted on NVIDIA GeForce RTX 3090 (24GB). Same GPU and package version are recommended.

Data Preparation

The public datasets employed in the paper are listed below. These dataset folders/files follow its original structure.

• MVTec AD (Download link)
• VisA (Download link)
• MPDD (Download link)

Run Experiments

For example, edit ./shell/run-mvtec.sh to configure arguments --datapath, --classes, and hyperparameter settings. Please modify argument --test to 'ckpt' / 'test' to toggle between training and test modes.

bash run-mvtec.sh

Citation

Please cite the following paper if the code help your project:

@article{chen2024progressive,
  title={Progressive Boundary Guided Anomaly Synthesis for Industrial Anomaly Detection},
  author={Chen, Qiyu and Luo, Huiyuan and Gao, Han and Lv, Chengkan and Zhang, Zhengtao},
  journal={IEEE Transactions on Circuits and Systems for Video Technology},
  year={2024},
  publisher={IEEE}
}

Acknowledgements

Thanks for the great inspiration from SimpleNet.

License

The code in this repository is licensed under the MIT license.