Research on Traffic Sign Detection Based on Convolutional Neural Network

Abstract

TSD (Traffic Sign Detection) is a hotspot in autonomous driving and assisted driving research. TSD research is of great significance for improving road traffic safety. In recent years, CNN (Convolutional Neural Networks) have achieved great success in object detecting tasks. It shows better accuracy or faster execution speed than traditional method. However, the execution speed and the detection accuracy of the existing CNN methods cannot be obtained at the same time. What's more, the hardware requirements are also higher than before, resulting in a larger detection cost. In order to solve these problems, this paper proposes an improved CNN model based on YOLO model, darknet 53 construction. By introducing batch normalization and RPN networks and improving the network structure for traffic sign detection tasks, the YOLO neural network detection model is optimized. The accuracy of the model in the traffic sign detection task is greatly improved, and the detection speed becomes faster. The results show that the method in this paper is of great help to improve the accuracy and detection speed of traffic sign detection and reduce the hardware requirements of the detection system as well.