Aperfeiçoamento de um sistema de reconhecimento de padrões do alfabeto de LIBRAS utilizando luva sensora

Abstract

The Brazilian Sign Language (LIBRAS) it’s a gestural-visual language used by deaf people in Brazil. In the teaching of LIBRAS, an interaction with the object of teaching can potentiate learning. From this premise, it is believed that a system of detection of the alphabet of LIBRAS can contribute in the learning process, as well as in the translation of the language into textual elements. Thus, this work presents the improvement of a system for recognizing patterns of the alphabet of LIBRAS using a previously conceived sensor glove. As a result, we obtained a system that uses neural networks capable of recognizing all the letters of the alphabet, with a correct percentage of 90,67%.