Anomaly-Guided Observation and Two-Level Semantic Feedback for Training-Free Video Anomaly Analysis

Key points

• Training-free language-based video anomaly detection (VAD) uses frozen vision--language models (VLMs) and large language models (LLMs) to localize abnormal events without task-specific model-weight optimization.1
• Existing approaches increasingly exploit broader temporal and semantic context, yet generic context does not necessarily foreground abnormal events or specify how video-level semantics should modify local anomaly scores.1
• Starting from an upstream anomaly-score curve, we adapt reasoning-based iterative frame selection: high-score regions seed the observations, and anomaly scores together with VLM relevance assessments guide subsequent temporal exploration.1
• On the HIVAU video-summary evaluations of UCF-Crime and XD-Violence, VAD-guided observation improves GPT-rated reasonability, detail, and consistency over the matched CLIP-guided variant.1

Sources (1)

• [1]Anomaly-Guided Observation and Two-Level Semantic Feedback for Training-Free Video Anomaly Analysis
  arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 12, 07:43 AM

  We introduce a training-free semantic feedback framework that uses VAD-guided active observation for video anomaly understanding (VAU) and returns the resulting anomaly-focused description to refine temporal detection.

  Training-free language-based video anomaly detection (VAD) uses frozen vision--language models (VLMs) and large language models (LLMs) to localize abnormal events without task-specific model-weight optimization.

Extractive summary: sentences quoted from the sources.