August 19, 2025Open Access

A Unified GAN-Based Framework for Unsupervised Video Anomaly Detection using Optical Flow and RGB Cues

View Full Paper

Ask AI

Bookmark

The framework achieves an Average Precision of 80.5% on XD-Violence, showcasing its superior performance compared to other unsupervised methods.
With an AUC of 0.92 and F1-score of 0.85 on Hockey Fight, it effectively detects short-duration violent events.
Using a unified GAN-based architecture, the model integrates RGB appearance and optical flow for enhanced detection accuracy.
DASLoss optimizes training by focusing on pixel, perceptual, temporal, and feature consistency, supporting effective generalization.

Discussion

No takes yet. Share an insight, caveat, or question.

Unsupervised framework improves video anomaly detection using optical flow and RGB cues, implying strong performance in diverse scenarios.

Kang et al. (2025) studied this question.

View Full Paper

Ask AI

Bookmark

Unsupervised framework improves video anomaly detection using optical flow and RGB cues, implying strong performance in diverse scenarios.