Neural Network Based on Multi-Scale Saliency Fusion for Traffic Signs Detection

Aiming at recognizing small-scale and complex traffic signs in the driving environment, a sign detection algorithm YOLO-FAM based on YOLOv5 is proposed. Firstly, new backbone network, ShuffleNet-v2, used to reduce algorithm’s parameters, realize lightweight detection, improve speed. Secondly, Bidirectional Feature Pyramid Network (BiFPN) structure introduced capture multi-scale context information, so as obtain more feature information accuracy. Finally, location added channel attention using Coordinated Attention (CA) mechanism, thus enhancing expression. The experimental results show that compared with YOLOv5, mAP value of this method increased by 2.27%. Our approach can be effectively applied scenes. At road intersections, planners better plan avoid jams.