Integration of Colbp and Viola Jones Feature Extraction Methods in Gender Classification Based on Facial Image

Abstract

Nowadays face recognition still being a hot topics to be discussed especially it’s utility for gender
classification. Gender classification is an easy task for human but it’s a challenging task for computers
because it doesn’t have capability for recognizing human gender without feature extraction. There are
still many researches about facial image feature extraction for gender classification. Geometry
features and Texture Features are well perform features for gender classification. This paper will
presents fusion of those feature in order to find an improvement for gender classifications task. Each
features will be extracted using Viola Jones Algorithm and Compass Local Binary Pattern method.
Both features will be combined using concatenated method in dataframe format. Viola Jones
algorithm has an issues when detecting each facial regions so it causes outliers in each geometry
features. The proposed method will be evaluated on color FERET dataset that contains facial images.
Classification task will be done using Random Forest and Backpropagation. The result is fusion
features present an improvement in gender classification using Backpropagation with 87% accuracy.
It prove that proposed method perform better than single feature extraction method.