Peruvian Sign Language Recognition Using Recurrent Neural Networks

• Autores: Barrientos-Villalta, Geraldine Fiorella, Quiroz, Piero, Ugarte, Willy
• Formato: artículo
• Fecha de Publicación: 2022
• Institución: Universidad Peruana de Ciencias Aplicadas
• Repositorio: UPC-Institucional
• Lenguaje: inglés
• OAI Identifier: oai:repositorioacademico.upc.edu.pe:10757/669597
• Enlace del recurso: http://hdl.handle.net/10757/669597
• Nivel de acceso: acceso embargado
• Materia: Deep learning; Recurrent neural networks; Sign language

Deaf people generally face difficulties in their daily lives when they try to communicate with hearing people, this is due to the lack of sign language knowledge in the country. Deaf people have to go on their everyday lives in company of a interpreter to be able to communicate, even wanting to go to buy bread every morning becomes a challenge for them and being treated in health centers also becomes a challenge, a challenge which should not exist since they have the fundamental right to health. For that reason this paper attempts to present a system for dynamic sign recognition for Peruvian Sign Language and our main goal is to detect which model and processing technique is the most appropriate to solve this problem. So that this system can be used in deaf people everyday life and help them communicate. There have been many projects around the world trying to address this situation. However, each Sign Language is unique in its own way and, therefore, a global and complete solution is not possible. There have also been similar projects in Peru, but all of them share the same flaw of only recognizing static signs. Since sign language is not just the static signs like the alphabet, a solution which addresses also words that can be used in sentences is needed. For this a dynamic recognition is needed, and this is the system that will be presented in this paper.