PROJECT TITLE :

 A Structure-Based Human Facial Age Estimation Framework under a Constrained Condition

ABSTRACT:

In computer vision and pattern recognition, developing an automatic age estimation approach for human faces continues to play a significant role. Many works on face age estimate concentrate on two aspects: extracting facial aging features and learning classification/regression models.

To distinguish our work from existing age estimation approaches, we investigate a novel aspect-system structure that is constrained: how to create a framework to increase age estimation performance based on the constraint given a fixed feature type and a fixed learning method? For facial age estimate, we present a four-stage fusion architecture. This framework begins with gender identification, then moves on to the second phase, gender-specific age grouping, the third stage, age estimation within age groups, and finally the fusion stage.

Three well-known benchmark datasets, MORPH-II, FG-NET, and CLAP2016, are used in the experiment to validate the approach. The experimental results show that adopting our suggested framework improves performance greatly, and that it also beats various state-of-the-art age estimation methods.