Pseudo Anomalies Are All You Need: Diffusion-Based Generation for Weakly-Supervised Video Anomaly Detection

TL;DR

This paper introduces PA-VAD, a diffusion-based weakly-supervised video anomaly detection method that synthesizes pseudo-abnormal videos from normal videos, achieving high accuracy without using real abnormal data.

Contribution

The paper proposes a novel generation-driven approach using diffusion models and CLIP-guided synthesis to train anomaly detectors without real abnormal videos.

Findings

Achieves 98.2% on ShanghaiTech dataset

Surpasses state-of-the-art on UCF-Crime dataset

Effective detection without real abnormal data

Abstract

Deploying video anomaly detection in practice is hampered by the scarcity and collection cost of real abnormal footage. We address this by training without any real abnormal videos while evaluating under the standard weakly supervised split, and we introduce PA-VAD, a generation-driven approach that learns a detector from synthesized pseudo-abnormal videos paired with real normal videos, using only a small set of real normal images to drive synthesis. For synthesis, we select class-relevant initial images with CLIP and refine textual prompts with a vision-language model to improve fidelity and scene consistency before invoking a video diffusion model. For training, we mitigate excessive spatiotemporal magnitude in synthesized anomalies by an domain-aligned regularized module that combines domain alignment and memory usage-aware updates. Extensive experiments show that our approach reaches 98.2% on ShanghaiTech and 82.5% on UCF-Crime, surpassing the strongest real-abnormal method on ShanghaiTech by +0.6% and outperforming the UVAD state-of-the-art on UCF-Crime by +1.9%. The results demonstrate that high-accuracy anomaly detection can be obtained without collecting real anomalies, providing a practical path toward scalable deployment.