Enhanced Deep Neural Network-Based Vehicle Detection System Using YOLOv11 for Autonomous Vehicles

This study presents an advanced vehicle detection framework based on YOLOv11, the latest evolution of the YOLO family of object detectors. The model is tailored for real-time autonomous driving scenarios, addressing longstanding challenges such as false detections, missed targets, and diverse environmental conditions. Building on previous YOLO iterations, YOLOv11 introduces architectural innovations, including the C3k2 block, SPPF module, and C2PSA units, resulting in superior precision, recall, and inference speeds. A dataset of 30,000 images, spanning six vehicle categories E-Scooter, bicycle, bus, car, motorcycle, and truck was used for training, validation, and testing. Rigorous evaluations yielded a precision of 97.2%, recall of 93.4%, mAP@50 of 97.0%, and mAP@50-95 of 86.6%. These metrics underscore YOLOv11’s robustness and adaptability to complex environments, ensuring consistently high detection accuracy and minimal misclassification. Ultimately, YOLOv11’s enhanced performance and efficiency position it as a promising solution for intelligent transportation systems (ITS), adaptive traffic control, and advanced driver assistance systems (ADAS).