Prominent Face Region Based Gender Classification Using Deep Learning
Sujit M. Deokar, Sanika S Patankar, Jayant V Kulkarni · 2018
The social interaction among people is always influenced by human face. The complex characteristics of human face are gender, emotions, identity, etc. In this world of visual technology, the challenge lies in automatic extraction of these facial characteristics. Hence research in this area has broad applications in biometric authentication, surveillance systems, human-computer interface, etc. This paper focuses on gender classification using face images. In proposed paper, deep learning based gender classification approach is used. In this paper, six prominent face regions are extracted using Haar classifier. These extracted face regions are then learned using deep learning architecture. The experimental results shows accuracy from 99.4% under various artifacts such as illumination variation, pose variation which are inherent in most of unconstrained environment. The database used is Psychological Image Collection at Stirling (PICS) 2D face database of Aberdeen.