Abstract: We propose a novel approach for video anomaly detection. Existing video anomaly detection methods train only on normal frames, with the expectation that the ...