Diseño e implementación de un prototipo para la traducción de lenguaje de señas mediante la utilización de un guante sensorizado.

Abstract

In this work it is proposed an intelligent system to translate sign language into text. The hardware of the prototype is formed by flex sensors, contact sensor and an inertial sensor, which are mounted on a polyester nylon glove. The software of the prototype consists of classification and machine learning algorithms: k-nearest neighbors, decision trees, and dynamic time warping. This system can learn to recognize static and dynamic gestures. Furthermore, it can learn to recognize a specific gesture of any user. This system was tested at translating 61 patterns (letters, numbers and some words) from the Ecuadorian sign language with 6 non-disabled people. Experimental results show an average recognition rate of 91.55%. These results show an important improvement with related works.