Fuzzy SVM for 3D facial expression classification using sequential forward feature selection

Facial expression detection is one of the emerging topics in computer vision. In this study, three-dimensional (3D) facial expression classification has been addressed. Firstly, a large set of features based on pair-wise distances of points in face model are extracted. The multi-class problem of fac...

Full description

Saved in:
Bibliographic Details
Published in2017 9th International Conference on Computational Intelligence and Communication Networks (CICN) pp. 131 - 134
Main Authors Zarbakhsh, Payam, Demirel, Hasan
Format Conference Proceeding
LanguageEnglish
Published IEEE 01.09.2017
Subjects
Bayes methods
Classification algorithms
Face
face model
facial expression
Feature extraction
feature selection
multi-class classification
Principal component analysis
Support vector machines
Three-dimensional displays
Online AccessGet full text

Cover

More Information
Summary:Facial expression detection is one of the emerging topics in computer vision. In this study, three-dimensional (3D) facial expression classification has been addressed. Firstly, a large set of features based on pair-wise distances of points in face model are extracted. The multi-class problem of facial expression detection is divided into 15 one-versus-one two-class classifiers. Sequential forward feature selection (SFFS) algorithm based on Naive Bayesian error rate is applied to select the most discriminative features. In the last step, a two level fuzzy SVM (FSVM) classifier is utilized in optimum low-dimensional feature space to detect multi-class labels of six basic expressions including anger, disgust, fear, happiness, surprise and sadness. Experiments conducted on BU-3DFE data set have proved that the performance of proposed algorithm is comparable with recent studies in this field.
ISSN:2472-7555
DOI:10.1109/CICN.2017.8319371