Automatic emotion classification: An approach using Eulerian Video Magnification and Artificial Neural Networks

Vitor Quaresma S. de Hollanda Ramos¹; Ana Paula G. S. de Almeida²; Carla M. C. C. e Koike¹; Dianne M. Viana²; Flavio de Barros Vidal¹

¹ Department of Computer Science, University of Brasilia; ² Department of Mechanical Engineering, University of Brasilia

Baixar PDF

Resumo

Researches using facial emotion detection, which involves making a computer being able to perceive the emotional state of a human user. In this work, we propose an initial approach to develop an automatic emotion classification using Eulerian Video Magnification and Artificial Neural Networks (ANN). The proposed approach creates a descriptor vector from processed Eulerian Magnification videos and uses an ANN classification to achieve a suitable accuracy rate for isolated and combined emotions.

Palavras-chave: Emotion recognition; Eulerian Magnification; Artificial Neural Network