Real-Time Pedestrian Detection and Tracking in Dynamic Urban Environments Using YOLO for Intelligent Video Systems

This paper describes a comprehensive system for real-time pedestrian detection and tracking, which is intended to suit the needs of intelligent video surveillance in dynamic metropolitan areas. The main detection component makes use of the most recent version of the YOLO (You Only Look Once) algorithm, which is noted for its speed and accuracy in identifying pedestrians in complicated settings. The CrowdHuman Pedestrain Dataset, a difficult dataset that replicates actual urban situations where a variety of environments, fluctuating lighting, and occlusions provide major obstacles, is used to benchmark the system. Our findings show that the system can be used for intelligent traffic management, public safety, and urban surveillance applications by improving both detection accuracy and tracking resilience while preserving real-time performance. This work contributes to the growing field of intelligent video analytics by combining cutting-edge deep learning techniques with practical tracking and interpretability solutions.