Sistema de detección de somnolencia en conductores de automóviles empleando técnicas de procesamiento de imágenes y Machine Learning.

Abstract

Artificial intelligence in recent years has gained considerable strength in relation to solving problems in real life and this has motivated how much, students, teachers, researchers, among others, are interested and become part of this topic as This is the case of the present proposal, where machine learning techniques are used. These techniques are applied in the classification of images to determine the degree of drowsiness in automobile drivers and, according to this result, issue a caution alert, determining in the same way which of them presents better performance. Among the techniques used we have: the k-Nearest Neighbor Algorithm (KNN), the Support vector machine algorithm (SVM) and a convolutional neural network (CNN). In addition, a mathematical development of some of the techniques used is carried out. The data set to be used is the so-called MRL EYE Dataset, this has collected the data of 37 different people (33 men and 4 women) and consists of 84,898 images, but only a percentage of these images is used. Apart from the images used from the database, classifications are made with images that do not belong to the database, but these images are taken by a webcam and the classification of images is also performed by implementing video in real time. At the end of this project, there is a robust and very useful tool in managing the issue of drowsiness in car drivers.