MIST: Multiple Instance Self-Training Framework for Video Anomaly Detection
Published in IEEE International Conference on Computer Vision and Pattern Recognition (CVPR), 2021
Recommended citation form:
Jia-Chang Feng, Fa-Ting Hong and Wei-Shi Zheng. “MIST: Multiple Instance Self-Training Framework for Video Anomaly Detection, Proceedings of the IEEE International Conference on Computer Vision and Pattern Recognition. 2021.
Weakly supervised video anomaly detection (WS-VAD) is to distinguish anomalies from normal events based on discriminative representations. Most existing works are limited in insufficient video representations. In this work, we develop a multiple instance self-training framework (MIST)to efficiently refine task-specific discriminative representations with only video-level annotations. In particular, MIST is composed of 1) a multiple instance pseudo label generator, which adapts a sparse continuous sampling strategy to produce more reliable clip-level pseudo labels, and 2) a self-guided attention boosted feature encoder that aims to automatically focus on anomalous regions in frames while extracting task-specific representations. Moreover, we adopt a self-training scheme to optimize both components and finally obtain a task-specific feature encoder. Extensive experiments on two public datasets demonstrate the efficacy of our method, and our method performs comparably to or even better than existing supervised and weakly supervised methods, specifically obtaining a frame-level AUC 94.83% on ShanghaiTech.
We have deployed extensive experiments on UCF-Crime and ShanghaiTech dataset, and outperforms other methods under the same setting.
Comparison with previous encoder-based method [Zhong .et al, CVPR 2019] as below, the left of which are the results of UCF-Crime while the right are those of ShanghaiTech.
Visualization of the testing results on UCF-Crime (better viewed in color). The red blocks in the graphs are temporal ground truths of anomalous events. The orange circle shows the wrongly labeled ground truth, the blue circle indicates the wrongly predicted clip, and the red cricle indicates the correctly predicted clip.
More spatial anomaly activation maps visualization on UCF-Crime. The left 5 columns of the graphs are the successful results while the right 2 columns are the failures.
The demo video is on Youtube, if you could not load it, you can clip here to watch the video.
The video is on Youtube, if you could not open it, you can clip here to watch the video.
It’s recommended to turn on the caption.
Paper Download paper here
Poster Download poster here