YOLO-Based Traffic Sign Detection for Autonomous Vehicles

Real-time traffic sign recognition is an essential part of intelligent transportation systems and autonomous vehicles. This paper describes a method involving YOLOv8n, a small object detection model, to conduct high-speed and accurate traffic sign recognition. The approach includes preprocessing, augmentation, and training on a labeled dataset of traffic signs. Our model results in a mean Average Precision (mAP) of 93.5, which beats earlier lightweight models in terms of efficiency. Our proposed method also proves to be resilient in the real world and thus can be deployed in embedded systems. Improved detection under conditions of extreme weather will be dealt with in the future.