Computational vision applied in automotive driving support systems

Abstract

The objectives of this work are: to build a system which recognizes traffic signs by analyzing the images/video taken with a camera installed in the car. The system includes three stages: Detection, Classification, and Recognition. The data set which I used to train and to test the model is German Traffic Sign Recognition Benchmark (GTSRB) data set [1], a publicly available data set for single-image. The system also used with Portuguese traffic signs database and several examples taken from Portuguese roads are used to demonstrate the effectiveness of the proposed system. Traffic signs are detected by analyzing color information, red and blue, and analyzing the shape of the signs as triangular, squared and circular shapes, contained in the images using Opencv library. To make the classifier, I used Convolutional Neural Network technique with TensorFlow as a Machine Learning framework. The recognition of traffic signs is done by comparing the data from classification phase with the ones of the database. The results in the classifier are almost 97,7 % [2], and results in detection part are 70% for red and blue traffic signs respectively.