%0 Conference Proceedings
%4 sid.inpe.br/sibgrapi/2016/09.10.01.56
%2 sid.inpe.br/sibgrapi/2016/09.10.01.56.17
%T Supervised Methods for Classifying Facial Emotions
%D 2016
%A Paiva, Francisco Aulísio dos Santos,
%A Costa, Paula Dornhofer Paro,
%A De Martino, José Mario,
%@affiliation Dept. of Computer Engineering and Industrial Automation, School of Electrical and Computer Engineering, University of Campinas (Unicamp)
%@affiliation Dept. of Computer Engineering and Industrial Automation, School of Electrical and Computer Engineering, University of Campinas (Unicamp)
%@affiliation Dept. of Computer Engineering and Industrial Automation, School of Electrical and Computer Engineering, University of Campinas (Unicamp)
%E Aliaga, Daniel G.,
%E Davis, Larry S.,
%E Farias, Ricardo C.,
%E Fernandes, Leandro A. F.,
%E Gibson, Stuart J.,
%E Giraldi, Gilson A.,
%E Gois, João Paulo,
%E Maciel, Anderson,
%E Menotti, David,
%E Miranda, Paulo A. V.,
%E Musse, Soraia,
%E Namikawa, Laercio,
%E Pamplona, Mauricio,
%E Papa, João Paulo,
%E Santos, Jefersson dos,
%E Schwartz, William Robson,
%E Thomaz, Carlos E.,
%B Conference on Graphics, Patterns and Images, 29 (SIBGRAPI)
%C São José dos Campos, SP, Brazil
%8 4-7 Oct. 2016
%I Sociedade Brasileira de Computação
%J Porto Alegre
%S Proceedings
%K Classification, emotions, facial expressions.
%X This paper presents a comparison between the K-NN (K-Nearest Neighbors) and SVM (Support Vector Machine) methods for classifying emotions. The database contains a set of 568 images of faces expressing 22 emotions. Classification is carried out in such a way as to classifying these 22 emotions as well as two other sets of categories, namely valence (positive and negative emotions) and the so-called six basic emotions (joy, sadness, fear, surprise, disgust, anger). Different sets of features were tested (statistics of histograms of regions of interest - mouth and eyes - and distances between characteristic points on the face) as well as different configurations of input parameters for training the classifiers in order to achieve the best performance. The results of the three experiments reveal accuracy values ranging from 79% to 90% for the K-.
%@language en
%3 2016_SIBGRAPI_WorkshopFace_CR.pdf