Mouse virtual conduzido através do rastreamento de movimentos da cabeça por webcam do computador

Abstract

Accessibility is a basic and indispensable condition for any social inclusion process. In Brazil, accessibility is a right supported by law that guarantees the safety and autonomy of people with disabilities or reduced mobility. Sometimes, when it comes to accessibility, an almost immediate relationship is made with architectural accessibility (ramps, handrails, etc.). However, digital accessibility has become a necessity of the first order, since access to information and the use of computers has intensified during this pandemic period of the new coronavirus in 2020. Unfortunately, people with disabilities still do not know many difficulties accessing the computer. This work proposes the development of a virtual mouse that, through the computer’s webcam, tracks the movements of the head and converts them into movements of the mouse cursor. It is also possible to simulate clicks with the detection of facial expressions (open eyes and high eyebrows). For the development of this work, he basically used the Python programming language (mainly an OpenCV library) and Convolutional Neural Networks. The results revealed that, despite the need for improvements, the virtual mouse driven through the movement of the head can help people with disabilities or reduced mobility of the upper limbs. It was observed 97.1 % of correctness in relation to the classification of images that would need to be classified as ”eyes closed”and 98.5 % of correctness with the class of ”high eyebrows”. In addition, it is important to emphasize that the proposal due to this work has the advantage of being of low cost, since it is not necessary to spend on any specific hardware, using only a common computer webcam.