TransAnomaly: Video Anomaly Detection Using Video Vision Transformer
·6645 words·32 mins
A prediction-based video anomaly detection approach combining U-Net and Video Vision Transformer (ViViT), with modifications for video prediction, capturing richer temporal and global context information, enabling anomaly localization.