Human face age estimation with adaptive hybrid features

This paper presents an appearance-based human face age estimation scheme. The age estimation has become an important study in recently several years. The main issue of the aging process is that it varies across various people, and which makes the age estimation rather challenge. This study combines the shape feature, texture feature, and frequency feature using Active Shape Model (ASM), Radon transform, and Discrete Cosine Transform (DCT) to establish robust adaptive hybrid features for further classification. In estimation stage, the SVM is employed for the proposed hierarchical classification framework and the SVR is also involved for regression. As documented in the experimental results, the proposed method can provide superior performance than former state-of-the-art methods in terms of the MAE with the FG-NET database.