Traffic Sign Detection Algorithm Based on Improved YOLOv5

Recently, driverless technology has developed rapidly, and improving the accuracy and speed of traffic sign detection is of great significance to reduce the frequency of traffic accidents and maintain the safety of people's lives and property. In order to solve the problem of low detection accuracy caused by small size, high density and complex background environment of traffic signs in actual production and life, an optimized YOLOv5s algorithm was proposed. First, SimAM was embedded in the Head part of YOLOv5s to adjust the attention weight, making the model more focused on important information and reducing the omission of small targets to a certain extent. Secondly, Focal GIoU was used to solve the problem of sample difficulty and class imbalance in target detection, so that the model could identify and locate the target more accurately. Finally, Mish was used as the activation function during model training, which improved the generalization ability and network optimization ability of the model. The experimental results show that the mAP.5 and mAP.5:.95 of the improved algorithm reach 93.8% and 87.8%, respectively. Compared with the base model, the improvement is increased by 3.4% and 3.5% respectively.