2511.13276

#1 Recognition of Abnormal Events in Surveillance Videos using Weakly Supervised Dual-Encoder Models [PDF¹] [Copy] [Kimi²] [REL]

Authors: Noam Tsfaty, Avishai Weizman, Liav Cohen, Moshe Tshuva, Yehudit Aperstein

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 transformer representations through top-k pooling, achieving 90.7% area under the curve (AUC) on the UCF-Crime dataset.

Subject: Computer Vision and Pattern Recognition

Publish: 2025-11-17 11:47:28 UTC

---