IMAGE CLASSIFICATION AND REGRESSION FOR GENDER AND AGE ESTI-MATION

Abstract

We introduce a deep learning method for automatic age and gender estimates from facial images.  Using a dataset fit for both classification and regression tasks, simultaneous prediction of gender (reaching 89% accuracy) and age (achieving 92% classification accuracy) was enabled.  This work advances the field by providing a strong deep-learning framework for facial analysis and by pointing out important elements like ideal network settings and data augmentation methods that affect the accuracy of age and gender estimates.