Traffic Sign Recognition Model Based on Spiking Neural Network

Abstract

Traffic sign recognition is one of the key technologies for intelligent transportation and automatic driving. Most of the existing recognition methods used Convolutional Neural Networks (CNN) to realize the breakthrough in accuracy. But CNN has problems such as high power consumption, large computational volume and slow speed in practical applications. Spiking Neural Network (SNN) is a deep learning structure based on simulating the mechanism of processing information in biological brain, which has stronger parallel processing capability, better sparsity and real-time performance. A traffic sign recognition model is designed in this paper. A traffic sign recognition model which based on spiking convolutional neural network incorporating the spatial attention mechanism is proposed (SA-SCNN); then an optimization method of image input coding is proposed to further improve the model recognition accuracy. Experiments show that the accuracy of the model proposed in this paper is 99.56% on the GTSRB.