Car Object Detection: Comparative Analysis of YOLOv9 and YOLOv10 Models

The field of computer vision known as car detection technology involves the identification and tracking of vehicles in images or video frames. Car detection is a widely utilized technology in various applications, including traffic control, parking management, and security systems. Recently, you YOLO (Only Look Once) based object detection methods have attracted significant attention in autonomous systems such as vehicle detection and parking systems due to their ability to perform real-time object detection. In this study, the performance of two state-of-the-art YOLO versions, YOLOv9 and YOLOv10, is evaluated using a vehicle image dataset. The experimental studies indicate that YOLOv10 has superior detection performance compared to YOLOv9. Additionally, both models have been demonstrated to achieve high accuracy in terms of training time and extraction speed. In particular, YOLOv10 demonstrated superior performance to YOLOv9, with recall ranging from 90% to 94% and precision rates between 94% and 96%. Furthermore, the YOLOv9 model exhibited faster training times and lower inference times, rendering it more suitable for real-time applications. Despite the challenges, including false positives and false negatives, our findings can contribute to the improvement of the accuracy and efficiency of car detection, thereby enhancing traffic control, parking management, and security systems.