Comparative Study of Vehicle Detection with Different YOLOv5 Algorithms

Vehicle sensing is key to implementing AI-based driving and monitoring systems. Vehicles on the road have increased dramatically. For that, managing the transportation system becomes difficult. To solve this problem, this article proposes a vision-based vehicle detection system. In this study, we developed real-time multi-object media detection based on “You only look once” algorithm (YOLOv5). We analyzed the accuracy of vehicle detection using YOLOv5s (small), YOLOv5n (nano), YOLOv5l (large), YOLOv5m (medium), and the largest of the five YOLOv5x. The test results confirm that the YOLOv5x model can provide higher detection accuracy than other algorithms. The main indicators of accuracy are Precision, Recall, and mAP (0.5).The determined accuracy of the YOLO5s, YOLOv5 m, YOLOv5n, YOLOv5l, and YOLOv5x algorithms on the dataset were 62.4, 64.2, 62.9, 68.7, and 69.7%. Our analysis shows that YOLOv5x is better and more efficient at detecting vehicles and can be implemented in real-time traffic control in traffic systems.