ReactVAU detects and explains video anomalies efficiently in live streams
ReactVAU: A Slow-Fast Decoupled Framework for Streaming Video Anomaly Understanding
Computer Vision and Pattern Recognition
Summary
Watching videos in real time to find unusual events is hard because many systems need to see the whole video before deciding, which isn’t possible live. The researchers created ReactVAU, a system that quickly spots strange things using a lightweight detector and saves important visual details so they aren’t lost over time. It only uses a more complex analysis when something suspicious happens, making it faster and better during long normal periods. Their tests show this method works well for finding and explaining odd events in live video feeds without slowing down.
video anomaly detectionstreaming videocausalityspatial grid foldingpersistent memorylarge multimodal language modelsreal-time processingsemantic reasoningsurveillance systemscomputational efficiency
Authors
Chia-Hui Chen, Shih-Ying Yeh, Fu-En Yang, Min-Hung Chen, Shang-Hong Lai
Abstract
In this paper, we propose ReactVAU, a Slow-Fast Decoupled Framework for real-time streaming Video Anomaly Understanding (VAU). Existing VAU methods rely on offline inference with global temporal sampling, which violates causality and prevents deployment in live surveillance streams. Conversely, general streaming video models satisfy causal access but dilute rare transient anomalies during memory compression and often invoke heavyweight MLLMs uniformly over long normal intervals. React VAU addresses this gap with three synergistic components: a lightweight Fast Detection Module based on Spatial Grid Folding (SGF) for continuous anomaly filtering; an Anomaly-Aware Persistent Memory (AAPM) that protects critical visual cues from temporal decay; and a heavyweight Slow Reasoning Module that remains dormant during normal streams and is awakened only by suspicious events for semantic verification and causal description. Extensive experiments on multiple benchmarks demonstrate that ReactVAU operates under strict streaming constraints while simultaneously achieving competitive performance in both anomaly detection and causal reasoning, alongside significantly enhanced computational efficiency by minimizing heavyweight MLLM invocations. Project page is available at https://huiyuiui.github.io/React_VAU/