Sign Language Recognition System Using Myo Armband and Neural Network

The interaction between a deaf person and a natural person has become a very recurring topic nowadays due to the need to ensure that these two groups of people can establish communication. This article proposes an intelligent system that allows interpreting the signs of the deaf language and converting them into written language so that a natural person can understand what a deaf person wants to express. This work makes a review of the literature in which it expresses the different methodologies and techniques that have been developed over time for the classification of deaf language signs, it also proposes a method that works in real time using the electromyography of the bracelet called "MYO bracelet" that recognizes 10 signs of deaf language through a sign collection, processing and classification algorithm. The characteristics obtained from the MYO are time and frequency statistics. For the word and phrase classification stage, this work proposes to use a multilayer perceptron artificial neural network. To verify the effectiveness of this work, an experiment was proposed with a group of 25 people for training and another 25 people for evaluation. The efficiency of the system was measured by the number of successes of the total number of people in the evaluation, obtaining an efficiency of 77% in the classification of signs.