Road sign recognition with Convolutional Neural Network

Extracting the contents of a digital image has been proven a hard problem for computers. Since for them, an image is only a matrix of values, knowing what structures a human would recognize in this image, is a nontrivial problem. In this paper, we have implemented and tested a system of detection of road signs. The approach taken in this work consists of using convolutional neural network where this network is supposed to distinguish between different classes of signs (stop, attention etc.) and the final model will then be integrated to the autonomous cars. Tests carried out on the dataset GTSRB (The German Traffic Sign Recognition Benchmark) shows the performance of the system currently being developed.