What is it about?
we propose a Particle filter-based framework for anomaly detection in the video. In general, we use particle filter to track the L2-norm computed from optical flow in video, whose initial particles distribution is extracted from the optical flow of normal video. The high residual in tracking will be regarded as an alert of abnormal events.
Featured Image
Photo by Jens Johnsson on Unsplash
Why is it important?
Our proposed method assumes that features from normal video are easy to predict while those from the abnormal ones are hard to. Experiments on UMN dataset show that our method can tackle specific anomalies. Our main contribution in this paper is the innovative application of PF to anomaly detection and the processing pipeline from video to alert signal
Perspectives
Our proposed method assumes that features from normal video are easy to predict while those from the abnormal ones are hard to. Experiments on UMN dataset show that our method can tackle specific anomalies. Our main contribution in this paper is the innovative application of PF to anomaly detection and the processing pipeline from video to alert signal
xw gao
Shanghai University
Read the Original
This page is a summary of: Particle Filter-based Prediction for Anomaly Detection in Automatic Surveillance, IEEE Access, January 2019, Institute of Electrical & Electronics Engineers (IEEE),
DOI: 10.1109/access.2019.2931820.
You can read the full text:
Contributors
The following have contributed to this page
