An Approach For Gender Classification

2480 Words Nov 25th, 2016 10 Pages
Abstract— In this paper, an approach for gender classification is carried out combining frontal face images, Haar cascades, Histogram of Oriented Gradients and Support Vector Machines. The comparison of the existing methods that delves into the effects of Haar Cascade Classifier and Histogram of Oriented Gradients(HOG) for Face Detection and the use of Support Vector Machines(SVM) for Gender Classification. A database of 2-D facial images was used, consisting of individual as well as group photographs. These images were used in face detection by Haar classifier and HOG and the results were compared at the end. Further, the detected faces were used to extract primary features on a face, such as, nose, eyes, lips, chin. Feature Extraction is used to calculate ratios, which differ for a Male and a Female. SVM uses these ratios in training and therefore classifying gender. Success ratio for HOG is found to be more than Haar Classifier and high accuracy is achievable with SVM.
Keywords— Face Detection, Gender Classification, Haar, HOG, SVM
1. INTRODUCTION
This paper presents novel approach to classifying gender from frontal facial images. Gender classification is one of the focuses of Human Computer Interaction (HCI) problem and has many potential applications. When we communicate with other people we process information about the person, such as the expression, gender, ethnicity and age. We hope human machine communication could flow more freely if the computer can comprehend…
Open Document