Real-time Crowd Monitoring and Management

Abstract

Efficient crowd monitoring is essential for public safety, space management, and emergency response planning. Traditional methods rely on manual observation, which is labour-intensive and prone to errors. This research presents a real-time crowd monitoring system that integrates classical computer vision techniques, enabling automated detection and density estimation of people in public spaces. The system employs background subtraction, contour detection, and optical flow tracking to monitor crowd movement dynamically. A heatmap visualization highlights high-density areas, providing insights for event organizers and security personnel. The solution is lightweight, runs on low-end hardware without GPU dependency, and delivers real-time analytics through a Flask-based web interface. Experimental results demonstrate the system's effectiveness in detecting and analyzing crowd behaviour, making it applicable in locations such as religious sites, transportation hubs, and public events.