Real-Time Anomaly Detection and Localization in Crowded Scenes
Abstract
In this paper, we propose a method for real-time anomaly detection and localization in crowded scenes. Each video is defined as a set of non-overlapping cubic patches, and is described using two local and global descriptors. These descriptors capture the video properties from different aspects. By incorporating simple and cost-effective Gaussian classifiers, we can distinguish normal activities and anomalies in videos. The local and global features are based on structure similarity between adjacent patches and the features learned in an unsupervised way, using a sparse auto-encoder. Experimental results show that our algorithm is comparable to a state-of-the-art procedure on UCSD ped2 and UMN benchmarks, but even more time-efficient. The experiments confirm that our system can reliably detect and localize anomalies as soon as they happen in a video.
Cite
Text
Sabokrou et al. "Real-Time Anomaly Detection and Localization in Crowded Scenes." IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops, 2015. doi:10.1109/CVPRW.2015.7301284Markdown
[Sabokrou et al. "Real-Time Anomaly Detection and Localization in Crowded Scenes." IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops, 2015.](https://mlanthology.org/cvprw/2015/sabokrou2015cvprw-realtime/) doi:10.1109/CVPRW.2015.7301284BibTeX
@inproceedings{sabokrou2015cvprw-realtime,
title = {{Real-Time Anomaly Detection and Localization in Crowded Scenes}},
author = {Sabokrou, Mohammad and Fathy, Mahmood and Hosseini, Mojtaba and Klette, Reinhard},
booktitle = {IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops},
year = {2015},
pages = {56-62},
doi = {10.1109/CVPRW.2015.7301284},
url = {https://mlanthology.org/cvprw/2015/sabokrou2015cvprw-realtime/}
}