TRACES: Temporal Recall with Contextual Embeddings for Real-Time Video Anomaly Detection

TL;DR

This paper introduces TRACES, a real-time, context-aware zero-shot video anomaly detection system that uses memory-augmented cross-attention to adaptively identify new anomalies with high accuracy and explainability.

Contribution

The work presents a novel memory-augmented, cross-attention based pipeline for real-time, zero-shot anomaly detection that significantly improves state-of-the-art performance.

Findings

Achieves 90.4% AUC on UCF-Crime

Achieves 83.67% AP on XD-Violence

Operates in real-time with high precision and explainability

Abstract

Video anomalies often depend on contextual information available and temporal evolution. Non-anomalous action in one context can be anomalous in some other context. Most anomaly detectors, however, do not notice this type of context, which seriously limits their capability to generalize to new, real-life situations. Our work addresses the context-aware zero-shot anomaly detection challenge, in which systems need to learn adaptively to detect new events by correlating temporal and appearance features with textual traces of memory in real time. Our approach defines a memory-augmented pipeline, correlating temporal signals with visual embeddings using cross-attention, and real-time zero-shot anomaly classification by contextual similarity scoring. We achieve 90.4\% AUC on UCF-Crime and 83.67\% AP on XD-Violence, a new state-of-the-art among zero-shot models. Our model achieves real-time inference with high precision and explainability for deployment. We show that, by fusing cross-attention temporal fusion and contextual memory, we achieve high fidelity anomaly detection, a step towards the applicability of zero-shot models in real-world surveillance and infrastructure monitoring.