Sistema basado en visión artificial para el reconocimiento de expresiones faciales aplicado a la enseñanza socioemocional en una asociación de autismo de Chiclayo
Cargando...
Fecha
2025
Autores
Piscoya Tirado, Jose Luis
Piscoya Tirado, Jose Luis
Título de la revista
ISSN de la revista
Título del volumen
Editor
Universidad Católica Santo Toribio de Mogrovejo
Resumen
Debido las dificultades que enfrentan los niños con Trastorno de Espectro Autista (TEA) para comunicar y reconocer emociones mediante expresiones faciales, el aumento global en la prevalencia del autismo (aproximadamente 1 de cada 100 infantes) y dado que el CONADIS registró 4528 casos a nivel nacional, resaltando que el 82% de casos en la región de Lambayeque corresponden a menores de edad. El objetivo general de esta investigación es crear un sistema basado en visión artificial para el reconocimiento de expresiones faciales aplicado a la enseñanza socioemocional de niños con TEA en una asociación de autismo de Chiclayo. Para ello, se diseñó un modelo de Redes Neuronales Convolucionales bajo la metodología CRISP-DM logrando un tiempo de respuesta de 40 milisegundos con una precisión y exactitud del 82%. Posteriormente, aplicando la metodología Kanban, se desarrolló un sistema de escritorio destinado a entrenar las habilidades de comunicación y reconocimiento emocional, empleando el modelo para clasificar expresiones en tiempo real a través de la cámara con una tasa de 12 fotogramas por segundo. Finalmente, se realizaron pruebas de usabilidad con 20 niños de 6 a 11 años con TEA de alto funcionamiento. La evaluación de aceptación tecnológica reveló que, entre los padres, el 78% y 76% están totalmente de acuerdo con la utilidad y facilidad de uso, respectivamente. Además, de los 6 profesionales de salud mental colaboradores, el 83% está totalmente de acuerdo en que la herramienta es útil, y el 75% la encuentra fácil de usar en las sesiones de enseñanza socioemocional.
Due to the difficulties that children with Autism Spectrum Disorder (ASD) face in communicating and recognizing emotions through facial expressions, the global increase in the prevalence of autism (approximately 1 in every 100 infants) and given that CONADIS registered 4,528 cases worldwide national, highlighting that 82% of cases in the Lambayeque region correspond to minors. The general objective of this research is to create a system based on artificial vision for the recognition of facial expressions applied to the socio-emotional teaching of children with ASD in an autism association in Chiclayo. To do this, a Convolutional Neural Networks model was designed under the CRISP-DM methodology, achieving a response time of 40 milliseconds with a precision and accuracy of 82%. Subsequently, applying the Kanban methodology, a desktop system was developed to train communication and emotional recognition skills, using the model to classify expressions in real time through the camera with a rate of 12 frames per second. Finally, usability tests were conducted with 20 children ages 6 to 11 with high-functioning ASD. The technological acceptance evaluation revealed that, among parents, 78% and 76% fully agree with the usefulness and ease of use, respectively. Additionally, of the 6 collaborating mental health professionals, 83% fully agree that the tool is useful, and 75% find it easy to use in social-emotional teaching sessions.
Due to the difficulties that children with Autism Spectrum Disorder (ASD) face in communicating and recognizing emotions through facial expressions, the global increase in the prevalence of autism (approximately 1 in every 100 infants) and given that CONADIS registered 4,528 cases worldwide national, highlighting that 82% of cases in the Lambayeque region correspond to minors. The general objective of this research is to create a system based on artificial vision for the recognition of facial expressions applied to the socio-emotional teaching of children with ASD in an autism association in Chiclayo. To do this, a Convolutional Neural Networks model was designed under the CRISP-DM methodology, achieving a response time of 40 milliseconds with a precision and accuracy of 82%. Subsequently, applying the Kanban methodology, a desktop system was developed to train communication and emotional recognition skills, using the model to classify expressions in real time through the camera with a rate of 12 frames per second. Finally, usability tests were conducted with 20 children ages 6 to 11 with high-functioning ASD. The technological acceptance evaluation revealed that, among parents, 78% and 76% fully agree with the usefulness and ease of use, respectively. Additionally, of the 6 collaborating mental health professionals, 83% fully agree that the tool is useful, and 75% find it easy to use in social-emotional teaching sessions.
Descripción
Palabras clave
Autismo, Reconocimiento de formas, Tecnología educativa, Autism, Pattern recognition, Educational technology
Citación
J. L. Piscoya Tirado. "Sistema basado en visión artificial para el reconocimiento de expresiones faciales aplicado a la enseñanza socioemocional en una asociación de autismo de Chiclayo," tesis de licenciatura, Fac. de Ingeniería, Univ. USAT, Chiclayo, Perú, 2023. [En línea]. Disponible en:
