SRVAU-R1: Enhancing Video Anomaly Understanding via Reflection-Aware Learning

TL;DR

This paper introduces SRVAU-R1, a reflection-aware learning framework that enhances video anomaly understanding by incorporating self-reflection and reasoning, leading to improved localization accuracy and reasoning quality.

Contribution

It presents the first reflection-oriented Chain-of-Thought dataset and a novel reflection-aware learning paradigm for better reasoning in video anomaly detection.

Findings

Outperforms existing methods on multiple benchmarks.

Achieves significant improvements in anomaly localization accuracy.

Enhances reasoning quality in video anomaly understanding.

Abstract

Multi-modal large language models (MLLMs) have demonstrated significant progress in reasoning capabilities and shown promising effectiveness in video anomaly understanding (VAU) tasks. However, existing MLLM-based approaches remain largely focused on surface-level descriptions of anomalies, lacking deep reasoning over abnormal behaviors like explicit self-reflection and self-correction. To address that, we propose Self-Reflection-Enhanced Reasoning for Video Anomaly Understanding (SRVAU-R1), a reflection-aware learning framework that incorporates reflection in MLLM reasoning. Specifically, SRVAU-R1 introduces the first reflection-oriented Chain-of-Thought dataset tailored for VAU, providing structured supervision with initial reasoning, self-reflection, and revised reasoning. Based on that, it includes a novel reflection-aware learning paradigm with supervised fine-tuning and reinforcement fine-tuning to enhance multi-modal reasoning for VAU. Extensive experiments on multiple video anomaly benchmarks demonstrate that SRVAU-R1 consistently outperforms existing methods, achieving significant improvements in both temporal anomaly localization accuracy and reasoning quality.