Development of traffic signs recognition WebService based on convolutional neural networks

Abstract

Image classification is one of the most important applications of neural networks. In this paper,we propose the classification algorithm for traffic signs recognition based on convolutional neural networks.The designed CNN is implemented using the TensorFlow framework, and the inference is performed using CUDA. To utilize the Connected Cars concept, we also developeda webservice to remotely process images, obtained by a camera installed into a vehicle. The experimental results show that the proposed algorithm, implemented using the Cuba library for developingclient-server apps, shows high efficiency and is applicable for a connected vehicle.