Comparative Analysis of People Counting and People Tracking Techniques in Smart Surveillance Systems

Smart monitoring systems improve city safety, retail measurement, and transport security and efficiency. These systems enhance public security, crowd control, and resource allocation by tracking individuals. People counting and tracking are two key characteristics of these systems, which identify and analyze human presence and movement patterns. People counting measures the number of people in certain regions, while people tracking tracks individuals over time and provides a more comprehensive behaviour analysis. People counting can be used to monitor mall foot traffic or improve public safety. In security surveillance, tracking individuals helps avoid accidents and other mishaps. This study evaluates the accuracy, computational complexity, and real-world applicability of these two approaches. The study compares human tracking and counting algorithms using the PETS 2009 and UCSD Pedestrian datasets in indoor and outdoor crowd environments with varying crowd densities. The results indicate that both approaches are practicable, with performance changing greatly depending on environmental conditions. People counting using YOLO is 95% accurate, while tracking with Deep SORT is 90%. Computational expenses (YOLO uses up to 70% CPU) and environmental change resistance remain issues, especially in real-time usage. These findings underscore the need for further study and improvement to enhance system performance in diverse surveillance settings.