Learning Causality-inspired Representation Consistency for Video Anomaly Detection

TL;DR

This paper introduces a causality-inspired framework for video anomaly detection that learns robust representations of normality by capturing underlying causal variables, improving detection accuracy and robustness against diverse regular event variations.

Contribution

It proposes a novel causality-inspired representation consistency framework that implicitly learns causal variables of normality directly from normal videos for improved anomaly detection.

Findings

Robust to label-independent shifts in normal events

Accurately detects complex anomalies in real-world videos

Outperforms existing methods in detection accuracy

Abstract

Video anomaly detection is an essential yet challenging task in the multimedia community, with promising applications in smart cities and secure communities. Existing methods attempt to learn abstract representations of regular events with statistical dependence to model the endogenous normality, which discriminates anomalies by measuring the deviations to the learned distribution. However, conventional representation learning is only a crude description of video normality and lacks an exploration of its underlying causality. The learned statistical dependence is unreliable for diverse regular events in the real world and may cause high false alarms due to overgeneralization. Inspired by causal representation learning, we think that there exists a causal variable capable of adequately representing the general patterns of regular events in which anomalies will present significant variations. Therefore, we design a causality-inspired representation consistency (CRC) framework to implicitly learn the unobservable causal variables of normality directly from available normal videos and detect abnormal events with the learned representation consistency. Extensive experiments show that the causality-inspired normality is robust to regular events with label-independent shifts, and the proposed CRC framework can quickly and accurately detect various complicated anomalies from real-world surveillance videos.