Crowd-Gathering Risk Management System Based on Collision Estimation

In this paper, we propose a crowd-gathering risk management system based on collision analysis. The system addresses the limitations of existing methods by not only detecting real-time crowd density but also predicting potential collision risks in densely populated areas. Our approach generates crowd grid maps, applies crowd clustering techniques, and predicts collision points and times using domain-specific data. The risk indicators produced are visualized to alert for potential hazards, ensuring effective crowd management and safety. We validated our system through simulations, demonstrating its ability to predict risks in high-density scenarios, making it a valuable tool for event organizers and public safety agencies.