3D Human-Human Interaction Anomaly Detection

TL;DR

This paper introduces a new task called Human-Human Interaction Anomaly Detection (H2IAD) and proposes IADNet, a novel neural network architecture that effectively captures complex interactive behaviors to detect anomalies in 3D human interactions.

Contribution

The paper presents the first framework specifically designed for detecting anomalies in collaborative 3D human interactions, incorporating novel modules for motion synchronization and social cue encoding.

Findings

IADNet outperforms existing baselines on human-human motion benchmarks.

The proposed modules effectively capture interaction dynamics and social cues.

The approach achieves higher accuracy in detecting interaction anomalies.

Abstract

Human-centric anomaly detection (AD) has been primarily studied to specify anomalous behaviors in a single person. However, as humans by nature tend to act in a collaborative manner, behavioral anomalies can also arise from human-human interactions. Detecting such anomalies using existing single-person AD models is prone to low accuracy, as these approaches are typically not designed to capture the complex and asymmetric dynamics of interactions. In this paper, we introduce a novel task, Human-Human Interaction Anomaly Detection (H2IAD), which aims to identify anomalous interactive behaviors within collaborative 3D human actions. To address H2IAD, we then propose Interaction Anomaly Detection Network (IADNet), which is formalized with a Temporal Attention Sharing Module (TASM). Specifically, in designing TASM, we share the encoded motion embeddings across both people such that collaborative motion correlations can be effectively synchronized. Moreover, we notice that in addition to temporal dynamics, human interactions are also characterized by spatial configurations between two people. We thus introduce a Distance-Based Relational Encoding Module (DREM) to better reflect social cues in H2IAD. The normalizing flow is eventually employed for anomaly scoring. Extensive experiments on human-human motion benchmarks demonstrate that IADNet outperforms existing Human-centric AD baselines in H2IAD.