A Hybrid Intelligent Crowd Monitoring System for Real-Time Public Safety

Abstract

Crowd monitoring has become an essential component of public safety management through continuous monitoring and detection of potential risks in congested areas. In this research, a hybrid intelligent crowd monitoring system comprising YOLOv8-based person detection, crowd density estimation, crowd flow analysis, abnormal crowd behavior detection, and alert generation is introduced. The design and implementation of the proposed hybrid intelligent crowd monitoring system have been accomplished utilizing the DMADV methodology in order to set up a defined design and implementation process. First of all, video frames from the surveillance cameras undergo pre-processing in order to improve their quality and then analyzed using YOLOv8 algorithm for person detection. Finally, the detected persons will be used for the estimation of crowd density and flow analysis as well as abnormal crowd behavior detection in real time. If the predetermined safety thresholds are exceeded, the automatic alert is generated.