Api rest para el reconocimiento facial de emociones (Fer Rest Api)
Archivos
Fecha
2019-06-26
Autores
López Mollinedo, Dairo
Título de la revista
ISSN de la revista
Título del volumen
Editor
Universidad Central “Marta Abreu” de Las Villas
Resumen
A nivel global existen muchas investigaciones que avanzan en la comprensión de las emociones para mejorar la experiencia de las personas al interactuar con una aplicación. Estos trabajos abarcan áreas muy diferentes entre sí, desde seguridad, salud, educación y robótica. En la Universidad Central “Marta Abreu” de Las Villas se creó una biblioteca para el reconocimiento facial de emociones llamada Feel UCLV. Para acceder desde otras aplicaciones a dicha biblioteca se necesita desarrollar un entorno de comunicación. Con el fin de simplificar la programación de métodos y funciones que permitan reconocer las emociones expresadas en el rostro humano para ser empleados en diferentes aplicaciones se decidió la creación de una API REST. Esta permite la detección automática de emociones faciales en tiempo real en aplicaciones de Computación Afectiva. Reconoce las seis emociones universales alegría, tristeza, miedo, ira, sorpresa y asco. Facilita el uso sin que el desarrollador tenga dominio de este tema. Está implementado en el lenguaje de programación Python mediante el framework FastAPI. Facial Emotions Recognition REST API incluye para la clasificación Red Neuronal Perceptrón Multicapa, Máquina de Soporte Vectorial, K Vecinos más Cercanos, Bosque Aleatorio, Árbol de Decisión, y Naive Bayes. Para evaluar los clasificadores utiliza los conjuntos de datos Conh-Kanade y KDEF alcanzando resultados satisfactorios.
Globally there is a lot of research that advances the understanding of the emotions to improve people’s experience when interacting with an app. Those works cover very different areas of each other, from safety, health, education and robotics. At the Universidad Central “Marta Abreu” de Las Villas a library was created for the facial recognition of emotions called Feel UCLV. To access such library from other applications a communication environment needs to be developed. In order to simplify the programming methods and functions that allow to recognize emotions expressed in the human face to be used in different applications it was decided to create a REST API. This allows automatic detection of facial emotions in real time in Affective Computing applications. It recognizes the six universal emotions happiness, sadness, fear, anger, surprise and disgust. Makes it easier to use without the developer having mastery of this topic. It is implemented in the Python programming language using the FastAPI framework. Facial Emotions Recognition REST API includes for the classification Multilayer Perceptron, Support Vectorial Machine, K Nearest Neighbor, Random Forest, Decision Tree, and Naïve Bayes. To evaluate the classifiers, use the Conh-Kanade and KDEF databases, achieving satisfactory results.
Globally there is a lot of research that advances the understanding of the emotions to improve people’s experience when interacting with an app. Those works cover very different areas of each other, from safety, health, education and robotics. At the Universidad Central “Marta Abreu” de Las Villas a library was created for the facial recognition of emotions called Feel UCLV. To access such library from other applications a communication environment needs to be developed. In order to simplify the programming methods and functions that allow to recognize emotions expressed in the human face to be used in different applications it was decided to create a REST API. This allows automatic detection of facial emotions in real time in Affective Computing applications. It recognizes the six universal emotions happiness, sadness, fear, anger, surprise and disgust. Makes it easier to use without the developer having mastery of this topic. It is implemented in the Python programming language using the FastAPI framework. Facial Emotions Recognition REST API includes for the classification Multilayer Perceptron, Support Vectorial Machine, K Nearest Neighbor, Random Forest, Decision Tree, and Naïve Bayes. To evaluate the classifiers, use the Conh-Kanade and KDEF databases, achieving satisfactory results.
Descripción
Palabras clave
Clasificadores, Computación Afectiva, Detección Automática, Emociones Faciales, API, Classifiers, Affective Computing, Automatic Detection, Facial Emotions, API