CADE: Continual Weakly-supervised Video Anomaly Detection with Ensembles

2025-12-07

AI Summary

To address catastrophic forgetting and anomaly under-detection in weakly supervised video anomaly detection (WVAD) caused by domain shift in open, dynamic environments, this paper pioneers the integration of continual learning (CL) into WVAD. We propose a Dual-Generator–Multi-Discriminator (DG-MD) architecture: dual generators mitigate label uncertainty and class imbalance inherent in weak supervision, while a multi-discriminator ensemble explicitly models and compensates for anomalous pattern degradation due to forgetting. Our method achieves continual adaptation without frame-level annotations by unifying generative adversarial training with model ensembling. Evaluated on multi-scenario benchmarks—including ShanghaiTech and Charlotte—our approach significantly outperforms state-of-the-art methods, effectively suppressing catastrophic forgetting and enhancing both cross-domain detection stability and completeness.

Technology Category

• Computer Vision: Adversarial Attacks & Robustness
• Machine Learning: Adversarial Learning & Robustness
• Multiagent Systems: Adversarial Agents

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Web Mining and Content Analysis: Robustness and generalizability of Web mining methods

Abstract

Video anomaly detection (VAD) has long been studied as a crucial problem in public security and crime prevention. In recent years, weakly-supervised VAD (WVAD) have attracted considerable attention due to their easy annotation process and promising research results. While existing WVAD methods tackle mainly on static datasets, the possibility that the domain of data can vary has been neglected. To adapt such domain-shift, the continual learning (CL) perspective is required because otherwise additional training only with new coming data could easily cause performance degradation for previous data, i.e., forgetting. Therefore, we propose a brand-new approach, called Continual Anomaly Detection with Ensembles (CADE) that is the first work combining CL and WVAD viewpoints. Specifically, CADE uses the Dual-Generator(DG) to address data imbalance and label uncertainty in WVAD. We also found that forgetting exacerbates the "incompleteness'' where the model becomes biased towards certain anomaly modes, leading to missed detections of various anomalies. To address this, we propose to ensemble Multi-Discriminator (MD) that capture missed anomalies in past scenes due to forgetting, using multiple models. Extensive experiments show that CADE significantly outperforms existing VAD methods on the common multi-scene VAD datasets, such as ShanghaiTech and Charlotte Anomaly datasets.

Problem

• Addresses domain-shift in video anomaly detection using continual learning
• Combats forgetting and data imbalance in weakly-supervised anomaly detection
• Ensembles models to detect missed anomalies from past scenes

Innovation

• Continual learning with ensembles for domain adaptation
• Dual-Generator addresses data imbalance and uncertainty
• Multi-Discriminator ensemble captures missed anomalies