A DNN based age and gender recognition technique for user identification during account creation in social media

Abstract

Social media is in the realms of our daily lives these days. From youngsters to aged and professionals to students – it is hard to find a person who does not have a social media account today. It is to be noted that social media today contains numerous fake accounts as there is no control or check of credentials at the time of opening the account. This uncontrolled expansion of social media has given rise to crimes; both physical and cyber. Criminals create fake account under masquerade and make friends with people to harm them physically, financially and mentally. In order to get rid of this, there should be a system of real time gender and age check before someone could open an account in social media. In this paper we have proposed a deep learning based method of age and gender detection in real time. The proposed method of age and gender detection may be used right at the time of opening an account. This way, it will be difficult for people to pose as someone, who they are not. In this paper we have made a comparative analysis of four main face detection algorithms based on Convolutional Neural Network namely Haar Cascade ,HOG , CNN and Deep Neural Network(DNN) to estimate an age-range and gender of a new face. The highest accuracy i.e 90 percent was achieved in case of DDN based algorithm for both age range approximation and gender detection.