Beyond Hazard Resemblance: Contrastive Event Adjudication for Training-Free Video Anomaly Detection

2026-08-10

AI Summary

This work addresses the limitations of existing training-free video anomaly detection methods, which rely on the semantic capabilities of pretrained models yet lack explicit criteria for distinguishing hazardous events from normal behaviors. To overcome this, we propose the Contrastive Event Adjudication for Video Anomaly Detection (CEAVAD) framework, which introduces—for the first time—a falsifiable hypothesis contrasting dangerous and benign events. During inference, CEAVAD dynamically constructs an anomaly boundary and leverages a pretrained vision-language model to perform evidence matching and contrastive reasoning, enabling both training-free anomaly detection and interpretable localization. The method achieves state-of-the-art performance under the training-free paradigm across three mainstream benchmarks.

Abstract

Video anomaly detection (VAD) aims to identify and temporally localize abnormal events in videos. Supervised methods learn anomaly decision boundaries from target-domain annotations but require substantial in-domain data. Existing training-free methods leverage the rich semantic knowledge and reasoning capabilities of pretrained models to interpret visual content, yet these capabilities do not directly define an anomaly decision criterion: richer anomaly descriptions better capture hazard resemblance without resolving abnormality. To this end, we propose Contrastive Event Adjudication for training-free Video Anomaly Detection (CEAVAD), which shifts the unit of inference from isolated anomaly concepts to falsifiable event hypotheses and establishes an inference-time explanatory boundary through the interaction between competing explanations and video evidence. Specifically, CEAVAD first uses public-safety knowledge to construct hazard-benign event contrasts, pairing each hazard mechanism with a generic normal account and a mechanism-specific benign counterpart. It then determines whether the target interval better supports a hazard explanation or its benign competitor, yielding a revisable contrastive boundary proposal for the target. Finally, CEAVAD adjudicates between the competing explanations to determine whether the hazard hypothesis survives the video evidence, supporting both temporally localized anomaly detection and evidence-grounded explanations. Experiments on three widely used VAD benchmarks demonstrate that CEAVAD achieves state-of-the-art performance under the training-free paradigm.

Problem

• video anomaly detection
• training-free
• anomaly decision criterion
• event adjudication
• contrastive reasoning

Innovation

• training-free
• contrastive event adjudication
• video anomaly detection
• explanatory boundary
• hazard-benign contrast