ComplexVAD: Detecting Interaction Anomalies in Video

2025-01-16

AI Summary

Existing video anomaly detection datasets lack explicit modeling of multi-object interaction anomalies, constraining methods to low-level pixel- or motion-based cues and limiting their ability to detect semantically complex interaction-level anomalies. To address this, we introduce ComplexVAD—the first large-scale benchmark dataset explicitly designed for interaction-aware anomaly detection—featuring systematic definition and fine-grained annotation of videos exhibiting inappropriate multi-object interactions (e.g., unauthorized collaboration, anomalous avoidance). We further propose a spatiotemporal scene graph modeling framework that unifies objects, actions, and spatiotemporal relations into dynamic graphs, and leverages graph neural networks with weakly supervised relational reasoning for semantic-level anomaly localization. Evaluated on ComplexVAD, our method significantly outperforms existing state-of-the-art approaches, demonstrating that explicit interaction modeling is critical for detecting complex, semantically grounded anomalies.

Technology Category

• Computer Vision: Video Understanding & Activity Analysis
• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning
• Reasoning under Uncertainty: Relational Probabilistic Models

Application Category

• Graph Algorithms and Modeling for the Web: Algorithms and analysis for heterogeneous, signed, attributed, multi-relational, temporal, higher-order, and annotated Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets

Abstract

Existing video anomaly detection datasets are inadequate for representing complex anomalies that occur due to the interactions between objects. The absence of complex anomalies in previous video anomaly detection datasets affects research by shifting the focus onto simple anomalies. To address this problem, we introduce a new large-scale dataset: ComplexVAD. In addition, we propose a novel method to detect complex anomalies via modeling the interactions between objects using a scene graph with spatio-temporal attributes. With our proposed method and two other state-of-the-art video anomaly detection methods, we obtain baseline scores on ComplexVAD and demonstrate that our new method outperforms existing works.

Problem

• Video Anomaly Detection
• Complex Interactions
• Dataset Limitations

Innovation

• ComplexVAD
• Scene Graphs
• Temporal-Spatial Attributes