Título: Otimização do Reconhecimento de Emoções com Ajuste de Hiperparâmetros e Balanceamento de Dados com Data Augmentation
Autores: Lara Toledo Cordeiro Ottoni, Jés de Jesus Fiais Cerqueira
Resumo: Facial expressions play a crucial role in non-verbal communication, influencing social interactions and collaboration between humans and robots. This study focuses on optimizing facial expression emotion recognition by exploring different combinations of optimizers, learning rates, data augmentation techniques, and neural network architectures. Using the FER2013 dataset, we evaluated the use of data augmentation for data balancing and six different CNN architectures. Our results indicate that the best-performing configuration includes the Adam optimizer with a learning rate of 0.001, the use of data augmentation, and a Base CNN architecture, achieving an accuracy of 77.83%.
Palavras-chave: Emotion recognition; Facial expressions; CNN; Deep learning.
Páginas: 7
Código DOI: 10.21528/CBIC2025-1191877
Artigo em PDF: CBIC_2025_paper1191877.pdf
Arquivo BibTeX:
CBIC_2025_1191877.bib