Video Abnormal Behavior Recognition and Hierarchical Warning System Based on Markov Model

Lingzi Jiang,Beiji Zou,Shu Liu,Yue Zhang,Zexin Ji
DOI: https://doi.org/10.1109/icccr56747.2023.10194153
2023-01-01
Abstract:In this paper, a video crowd anomaly detection and classification method based on Markov model is proposed. The anomaly detection in crowd is defined as different abnormal targets in dense crowd, and on this basis, a human feature model or a sample describing human features is established. In order to verify the effectiveness of this method, ShanghaiTech, UMN and UCSD human behavior database are used to analyze and verify the performance of the proposed Markov model feature extraction algorithm. The experimental results of video abnormal behavior recognition and detection in the above three databases using the algorithm proposed in this paper show that: The recognition rates of ShanghaiTech, UMN and UCSD human behavior databases are 94.65%, 93.46% and 90.31%.
What problem does this paper attempt to address?