Classification of gender from face images and voice

Abstract

Automated acknowledgement of a human gender is pivotal for various frameworks like data recovery, human–machine collaboration that processes human source data. Automatic gender classification has become relevant to an increasing amount of applications, particularly since the rise of social platforms and social media. Gender identification has been a widely researched area, but face attribute recognition from facial images still remains challengeable. We propose a methodology for automatic gender classification based on feature extraction from facial images. Our methodology includes three main iterations: preprocessing, feature extraction, and classification. Texture features, geometric moments, and histogram values of facial images are used to train the system. The eminent and efficient features are selected and trained using suitable machine learning technique classifiers, namely SVM, AdaBoost and random forest. The developed model is used to identify the gender of an individual with an accuracy of approximately 95% and can be deployed in various scenarios like hospital registration process, sports selection process, and airports to identify the gender of a person.