An Integrated AI and IoT-Based Intelligent Crowd Control and Management Framework for Large-Scale Events

Abstract

Large public gatherings at locations such as railway stations, temples, shopping centers, and event venues have become increasingly common, leading to heightened risks associated with overcrowding and stampede incidents. Many of these environments still lack affordable and real-time crowd monitoring solutions capable of supporting timely preventive action. To address this challenge, this work presents a cost- effective and scalable intelligent crowd monitoring system that operates using a mobile phone configured as an IP camera and a laptop for local, on-device processing. The proposed system employs the YOLOv8 deep learning model for real- time human detection, integrated with object tracking methods to enable accurate crowd counting, along with pretrained convolutional neural networks for basic demographic analysis. A lightweight Flask-based web interface is used to visualize live crowd statistics and automatically generate alerts when predefined crowd thresholds are exceeded. Experimental results indicate that the system is able to deliver reliable real-time performance on low-cost hardware, demonstrating the feasibility of edge-based AI solutions for enhancing public safety and supporting smart city applications.