Reconocimiento de emociones faciales utilizando Deep Learning
Loading...
Official URL
Full text at PDC
Publication date
2024
Advisors (or tutors)
Editors
Journal Title
Journal ISSN
Volume Title
Publisher
Citation
Abstract
El reconocimiento de expresiones faciales es una tarea crucial en el campo de la visión artificial, fundamental en la interacción persona-ordenador. En los últimos años, se ha demostrado la eficiencia de las redes neuronales para llevar a cabo esta tarea. En este trabajo, presentamos una red neuronal convolucional para el reconocimiento de expresiones faciales en fotografías de seres humanos en situaciones cotidianas. Asimismo, profundizamos en diferentes técnicas para mejorar el aprendizaje de nuestro modelo, empleando nuevos algoritmos no aplicados en este problema hasta ahora. Para ello, comenzamos introduciendo varias iteraciones de una arquitectura de red neuronal convolucional, con ligeras variaciones entre sí. A continuación, entrenamos cada una de ellas aplicando técnicas de preprocesamiento y de aumento de datos para conseguir modelos con una alta capacidad clasificativa. Finalmente, demostramos la precisión general de nuestro modelo y lo comparamos con los modelos más efectivos en la actualidad.
Facial expression recognition is a crucial task in the field of computer vision, fundamental for human-computer interaction. Recent years have witnessed and increased efficiency of neuronal networks to perform this task. Here, we present a deep convolutional neural network for facial expression recognition in photographs of humans in everyday situations. We also delve into different techniques to improve the learning of our model, making use of new algorithms not applied in this problem so far. To do so, we start by introducing several iterations of a convolutional neural network architecture, with slight variations. We then train each of them by applying preprocessing and data augmentation techniques to achieve models with high classification capability. We demonstrate the overall accuracy of our model and compare it with current most effective models.
Facial expression recognition is a crucial task in the field of computer vision, fundamental for human-computer interaction. Recent years have witnessed and increased efficiency of neuronal networks to perform this task. Here, we present a deep convolutional neural network for facial expression recognition in photographs of humans in everyday situations. We also delve into different techniques to improve the learning of our model, making use of new algorithms not applied in this problem so far. To do so, we start by introducing several iterations of a convolutional neural network architecture, with slight variations. We then train each of them by applying preprocessing and data augmentation techniques to achieve models with high classification capability. We demonstrate the overall accuracy of our model and compare it with current most effective models.
Description
Trabajo de Fin de Grado en Ingeniería Informática, Facultad de Informática UCM, Departamento de Sistemas Informáticos y Computación, Curso 2023/2024.