Regresar

Ecuadorian traffic sign detection through color information and a convolutional neural network

Abstract:

This article develops an algorithm for the detection and recognition of regulatory and warning traffic signs of Ecuador. The method consists of the following steps, i) video stabilization to reduce vertical oscilation, ii) implementation of a new method based on color segmentation and a convolutional neural network. Another important contribution is a new database of regulatory and prevention traffic signs of Ecuador. This dataset is with more than 37500 images of Ecuadorian road signs divided in 52 classes; these images were taken from different angles and weather conditions. The final version of this proposal runs at 22 frames per second, and it was tested in real driving conditions, in several roads and highways of Ecuador, during the day.