Lightning Fast Video Anomaly Detection via Adversarial Knowledge Distillation
This work addresses the need for real-time anomaly detection in video surveillance by providing a significantly faster method, though it is incremental as it builds on existing distillation techniques.
The paper tackles the problem of slow video anomaly detection by proposing a frame-level model that uses adversarial knowledge distillation from multiple object-level teachers, achieving a speed of 1480 FPS, which is over 7 times faster than the fastest competing method, with comparable accuracy on benchmarks like Avenue, ShanghaiTech, and UCSD Ped2.
We propose a very fast frame-level model for anomaly detection in video, which learns to detect anomalies by distilling knowledge from multiple highly accurate object-level teacher models. To improve the fidelity of our student, we distill the low-resolution anomaly maps of the teachers by jointly applying standard and adversarial distillation, introducing an adversarial discriminator for each teacher to distinguish between target and generated anomaly maps. We conduct experiments on three benchmarks (Avenue, ShanghaiTech, UCSD Ped2), showing that our method is over 7 times faster than the fastest competing method, and between 28 and 62 times faster than object-centric models, while obtaining comparable results to recent methods. Our evaluation also indicates that our model achieves the best trade-off between speed and accuracy, due to its previously unheard-of speed of 1480 FPS. In addition, we carry out a comprehensive ablation study to justify our architectural design choices. Our code is freely available at: https://github.com/ristea/fast-aed.