README.md

---
license: mit
pipeline_tag: video-classification
---

# FG-Diff: Frequency-Guided Diffusion Model with Perturbation Training for Skeleton-Based Video Anomaly Detection

This repository contains the pre-trained checkpoints and code for the paper [Frequency-Guided Diffusion Model with Perturbation Training for Skeleton-Based Video Anomaly Detection](https://huggingface.co/papers/2412.03044).

- **Project page:** [https://xiaofeng-tan.github.io/projects/FG-Diff/index.html](https://xiaofeng-tan.github.io/projects/FG-Diff/index.html)
- **Code:** [https://github.com/xiaofeng-tan/fgdmad-code](https://github.com/xiaofeng-tan/fgdmad-code)

## Overview

FG-Diff is a frequency-guided diffusion model for skeleton-based video anomaly detection. It improves robustness in open-set scenarios through perturbation training and uses frequency information to focus on principal motion components.

## Checkpoints

Pre-trained models are available on Hugging Face: [https://huggingface.co/ModelsWeights/AD-FG-Diff](https://huggingface.co/ModelsWeights/AD-FG-Diff)

## Usage

Please refer to the [GitHub repository](https://github.com/xiaofeng-tan/fgdmad-code) for detailed setup, training, and evaluation instructions.

## Citation

If you find this work useful, please consider citing:

```bibtex
@article{tan2026fgdiff,
  title={Frequency-Guided Diffusion Model with Perturbation Training for Skeleton-Based Video Anomaly Detection},
  author={Tan, Xiaofeng and Wang, Hongsong and Geng, Xin and Wang, Liang},
  journal={IEEE Transactions on Image Processing},
  year={2026},
  doi={10.1109/TIP.2026.3730816}
}
```