A High-Precision Road Sign Recognition Method Based on Improved YOLOv5s
Chaoliang Zhong, Riqian Hu, Yao Zhou · 2023
An improved traffic sign detection model based on YOLOv5s is proposed to address the problem of low accuracy of traffic sign detection in real road scenarios. The model introduces Concat connections in the SPP module to reduce the loss of feature map information and improve the feature representation capability; at the same time, the model introduces a coordinate attention mechanism in the backbone network to increase the perceptual domain of the network and improve the detection accuracy of traffic sign recognition. Experimental analysis of the CCTSDB dataset shows that the average accuracy of the improved YOLOv5s model is 91.78%, which is 10.4% higher than the original YOLOv5s. The improved YOLOv5s model can effectively identify small targets, dense and obscured traffic signs and has better detection stability. It is more suitable for high-speed traffic sign recognition in complex environments.