1 paper
Noam Tsfaty, Avishai Weizman, Liav Cohen +2
We address the challenge of detecting rare and diverse anomalies in surveillance videos using only video-level supervision. Our dual-backbone framework combines convolutional and t…