Age/gender classification with whole-component convolutional neural networks (WC-CNN)

A novel Convolutional Neural Network (CNN) solution is presented in this work to achieve human age/gender classification from their facial image data in an unconstrained environment. Our proposed Whole-Component CNN solution contains both the whole face and the facial component networks, and a confidence analysis module to handle the classification task. The proposed WC-CNN solution outperforms prior art in the challenging Adience database by a significant margin.