A Traffic Sign Recognition Method Based on Improved VGGNet Model

Traffic sign recognition is one of the key technologies in auto drive system. In recent years, with the wide application of convolutional neural network, the recognition effect has made great progress. However, the current popular networks still have shortcomings in lightweight and recognition accuracy. An improved Visual Geometry Group network (VGGNet) model is proposed in this paper based on the VGGNet, which achieves the goal of extracting fine features and network lightweight by adjusting the structure and number of convolutional layers. The proposed network was trained and tested on the German traffic sign database, and the recognition accuracy reached 98.8%.