<![CDATA[Human gender could be recognized from his or her face. Males and females have many different features, such as face shape, eyebrows, mouth, chin, nose, eyes, including facial hairs. Many fields aided by such system which could be developed to automatically recognize human gender, especially for demographic analysis by Indonesian Government. Such gender classification system will rapidly help decision maker that need gender recognition ability. A classifier model could be built to distinguish males from females from its facial features by learning a collections of male and female images data. One method for shape feature extraction is Histogram of Oriented Gradients (HOG). Results shows that classifier ability could be improved by tuning HOG parameter like size for dividing to local image regions, orientation histogram bin size and how each histogram relate to another. This research discussing case of subjects wearing glasses and not. This research explains how to build classification model from Histogram of Oriented Gradients based on face images. Built model able to classify men and women up to 97,83% and 95,92% each. Best parameter for Histogram of Oriented Gradients to classify gender is using (8,8) pixels per cell, 9 bin histogram each cell, (2,2) cell per blocks from (128,128) face image. It could be concluded too that glasses shape affects classification model ability.]]>
Copyrights © 2019
