Comparação de Optimizadores de Deep Learning em Reconhecimento de Expressões Faciais

• Autor(a) principal: Alves, Miguel António Madureira Fontoura
• Data de Publicação: 2022
• Tipo de documento: Dissertação
• Idioma: eng
• Título da fonte: Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
• Texto Completo: http://hdl.handle.net/10400.22/22233
• topic: Artificial Intelligence; Motion; FER; Deep Learning; Optimizer; Inteligência Artificial; Emoção; Reconhecimento de Expressão Facial; Otimizador

Artificial Intelligence is everywhere we go, whether it is programming an interactive cleaning robot or detecting a bank fraud. Its rise is inevitable. In the last few decades, many new architectures and approaches were brought up, so it becomes hard to know what is the best approach or architecture for a certain area. One of such areas is the detection of emotion in the human face, most commonly known by Facial Expression Recognition (or FER). In this work we started by doing an intensive collection of data concerning the theories that explain the existence of emotions, how they are distinguished from one another, and how they are recognized in a human face. After this, we started to develop deep learning models with different architectures as to compare their performances when used for Facial Expression Recognition. After developing the models, we took one of them and tested it with different deep learning optimizer algorithms, as to verify the difference among them, thus figuring out the best optimizing algorithm for this particular case.