Shearlet Convolutional Neural Network Approach for Age and Gender recognition

Automatic age and gender classification has become a very important field today, which is used in several applications (e.g. social networks). The key factors that influence clearly the age and gender recognition rate are the Feature Extraction Function (FEF) and the classification method. The need to further improve this rate continues to increase day by day. To try to further enhance the recognition rate, we propose in this contribution an approach based on: Discrete Shearlet Transform (DST) us a first step of manual feature extraction layer, and Deep Convolutional Neural Network (DCNN) us a second automatic feature extraction layer and classification step. The ability of the shearlet to extract significant features and of the neural networks to classify input data, will allow us to improve the age/gender recognition rate.