Method to Detect Movement in the Eyes Using Neural Networks. Peru Case

Abstract

The detection of eye movements is a promising technology for the management and control of devices, which offers innumerable possibilities to people with certain physical limitations. In today's society there is a gap that needs to be overcome and obstacles that need to be removed in favor of people with certain physical disabilities, in order to allow them to carry out activities in conditions of equality with others. There are many people with problems of paralysis in the upper and lower extremities which prevent them from being able to carry out their most basic activities by their own means, depending for this purpose on the help of third parties. The purpose of this research was to develop a method, which allowed the detection of eye movement, making use of artificial neural networks. This method consists of four stages, which are: Cataloging of the physical movements of the eyes; Image processing; neural network design and training; and finally the implementation of a prototype and performance measurement of the neural network. This research is considered very important in trying to achieve a method that is economical and efficient, the same that allows the detection of eye movements, which serve for further research that aims to guide mobile vehicles or control of devices with organs other than the extremities, to be more exact with the movement of the eyes, that is, through the gaze. An environment has been developed in PHP as a prototype or test environment, the evaluation of the performance of the neural network, as well as for obtaining the characteristic matrices of the dataset, which are taken from photographic images of the human eye, detecting the position of the iris, and according to this the movement will be determined. The training of the artificial neural networks was carried out in the Joone framework. © 2023 Elsevier B.V., All rights reserved.