An Effective Age Estimation Method using Scattering Transform

Omaima FathElrahman Osman

Age estimation, scattering transform, texture feature , support vector regression.

With the increasing demand of automatic recognition and surveillance systems, research on human faces such as face recognition, face detection, facial expression recognition and gender classification, have attracted significant attention in both computer vision and pattern recognition. Compared to these faces related research, face age estimation is one of the popular research topics. It is an important technique that used in many real-world applications, such as security control and monitoring, soft biometrics and Electronic Customer Relationship Management (ECRM). In this paper a new method for age estimation was proposed.  Where an efficient descriptor, scattering transform, which scatters the Gabor coefficients and pooled with Gaussian smoothing in multiple layers, is evaluated for facial feature extraction.  Then support vector regression was used for age prediction. The experimental results show that scattering transform-based descriptor yielded a superior result on benchmark datasets FERET and PAL, with MAE 2.12 and 2.22 respectively.