REAL-TIME VEHICLE DETECTION AND TRACKING USING YOLOV5 AND DEEPSORT FOR INTELLIGENT TRANSPORTATION SYSTEMS

The rapid urbanization and increasing number of vehicles have led to significant challenges in traffic management, including congestion, safety issues, and environmental pollution. Intelligent Transportation Systems (ITS) require reliable and efficient methods for real-time vehicle detection and tracking. This paper presents a comprehensive analysis of a vehicle detection and tracking system based on YOLOv5 and DeepSORT algorithms. The proposed architecture consists of three main stages: video frame extraction, object detection using YOLOv5 (with its backbone, neck, and head components), and multi-object tracking using DeepSORT (incorporating Kalman filter, Hungarian algorithm, and Re-ID features). The system outputs annotated video with bounding boxes, unique ID numbers, speed estimation, and traffic density calculation. Experimental results demonstrate that the proposed approach achieves 87.93% detection accuracy and effectively maintains consistent vehicle identities under various traffic conditions. The system shows promising potential for real-time traffic monitoring applications in smart cities.