Reconocimiento de gestos mano en tiempo real utilizando aprendizaje automático e información infrarroja.

Abstract

Hand gesture recognition systems are widely used in other disciplines such as medicine, virtual reality, augmented reality, robotics and others. In the field of robotics, it improves human computer interaction, in the field of medicine it brings a lot of value in working in sterile rooms, while virtual and augmented reality make a great contribution to education. However, hand gesture recognition is not a trivial problem, it is considered as a problem related to pattern recognition and feature extraction. This thesis presents the study of 6 feature selection methods and 17 feature extraction functions. These methods have been evaluated using machine learning algorithms such as: Decision Trees (DT), k-Nearest Neighbors (k-NN), Support Vector Machines (SVM) and Artificial Neural Networks (ANN). In addition, models are developed and evaluated using both manual and automatic feature extraction. Manual feature extraction models are implemented and analyzed using the aforementioned algorithms, achieving a test accuracy of 92.936%. While automatic feature extraction models are developed using 1D Convolutional Neural Networks (CNN) and Bi-directional Long Short Term Memory Neural Networks (BiLSTM), achieving an accuracy of 99.999%.