Software traductor de palabras básicas de lengua de señas colombiana
No Thumbnail Available
Date
2024
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Institución Universitaria Pascual Bravo
Abstract
La comunicación entre personas sordas y oyentes presenta barreras debido a la falta de conocimiento generalizado sobre la lengua de señas. Este proyecto aborda esta problemática mediante el desarrollo de un software capaz de reconocer y traducir en tiempo real siete palabras (hola, adiós, gracias, mamá, papá, sordo, casa) de la lengua de señas colombiana (LSC), usando redes neuronales convolucionales. El objetivo es facilitar la interacción entre personas sordas y oyentes, especialmente en situaciones cotidianas, promoviendo así una mayor inclusión social. La intervención realizada incluyó la recopilación de un conjunto de imágenes que capturan gestos específicos en LSC, y el entrenamiento de un modelo de inteligencia artificial en Python mediante las bibliotecas TensorFlow y Keras. A partir de pruebas de desempeño realizadas el modelo provee una precisión del 81% a 95% en el reconocimiento de las siete palabras.
-- Abstract Communication between deaf and hearing individuals significant barriers due to the limited knowledge of sign language among the general population. This project addresses this issue by developing software capable of recognizing and translating seven key words in Colombian Sign Language (LSC)—hello, goodbye, thank you, mom, dad, deaf, and home—in real time using convolutional neural networks. The objective is to facilitate interaction between deaf and hearing people, particularly in everyday situations, thus promoting greater social inclusion. The project involved compiling a set of images capturing specific gestures in LSC and training an artificial intelligence model in Python utilizing TensorFlow and Keras libraries. Performance tests indicate that the model achieves an accuracy rate between 81% and 95% in recognizing the selected words.
-- Abstract Communication between deaf and hearing individuals significant barriers due to the limited knowledge of sign language among the general population. This project addresses this issue by developing software capable of recognizing and translating seven key words in Colombian Sign Language (LSC)—hello, goodbye, thank you, mom, dad, deaf, and home—in real time using convolutional neural networks. The objective is to facilitate interaction between deaf and hearing people, particularly in everyday situations, thus promoting greater social inclusion. The project involved compiling a set of images capturing specific gestures in LSC and training an artificial intelligence model in Python utilizing TensorFlow and Keras libraries. Performance tests indicate that the model achieves an accuracy rate between 81% and 95% in recognizing the selected words.