Enhanced Crowd Analysis Using YOLO

Crowd analysis plays a vital role in public safety, event management, and disaster prevention by enabling real-time monitoring and anomaly detection in densely populated areas. Deep learning-based object detection models, particularly the You Only Look Once (YOLO) family, have demonstrated remarkable efficiency in such applications. In this paper, we perform a comparative analysis of YOLOv3, YOLOv5, YOLOv7, and YOLOv8 for crowd analysis, evaluating their performance in terms of detection accuracy, inference speed, model complexity, and suitability for real-time deployment. Experiments are conducted on datasets representing varying crowd densities and occlusion levels to assess each model’s effectiveness in detecting individuals in complex environments. The results highlight the trade- offs between accuracy and computational efficiency, pro- viding insight into the optimal choice of YOLO models for different crowd-analysis scenarios.