Traffic Sign Detection in Strong Glare Scenes

A traffic sign detection model based on image enhancement and the GD-YOLOv8 algorithm is proposed to address the issue of low detection accuracy of existing traffic sign detection models in strong glare scenes. For low-quality traffic sign images, histogram equalization is adopted to enhance overexposed image regions. To address the issues of missed detections and false alarms, a traffic sign detection model based on the GD-YOLOv8 algorithm is designed to fully extract traffic sign feature information. Experimental results on the GLARE dataset show that the proposed method achieves an average improvement of 12.31% in detection accuracy compared to SSD, Faster R-CNN, YOLOv5s, YOLOv7, and YOLOv8n, demonstrating its effectiveness in strong glare scenes.