Real-Time Traffic Monitoring and Analysis Using YOLO-Based Object Detection

In the context of Intelligent Transportation Systems (ITS), the rapid and precise detection, classification, and speed estimation of vehicles play a critical role in enhancing traffic management and law enforcement. Challenges such as minimal inter-vehicular spacing and visual noise in video frames can hinder accurate vehicle identification. Leveraging the proliferation of surveillance infrastructure in urban environments, this study introduces a novel framework that combines the YOLOv7 object detection algorithm with Kalman filtering to achieve real-time, non-intrusive vehicle monitoring. YOLOv7, known for its high-speed performance and accuracy, is employed to detect and classify vehicles from video streams, achieving processing rates up to 155 frames per second. To ensure consistent tracking and estimate vehicular speed, a Kalman Filter-based approach is integrated, allowing for dynamic prediction of vehicle trajectories. The methodology also incorporates a centroid-based tracking mechanism for vehicle counting along predefined paths. The proposed system effectively addresses the limitations of conventional physical sensors and demonstrates significant potential for real-time traffic regulation enforcement and data-driven transport analytics.