Video anomaly detection (VAD) – commonly formulated as a multiple-instance learning problem in a weakly-supervised manner due to its labor-intensive nature – is a challenging problem in video ...