Automatic Age Estimation from Face Images via Deep Ranking
Journal
26th British Machine Vision Conference, BMVC 2015
Start Page
55.1
End Page
55.11
Date Issued
2015-09
Author(s)
Abstract
Automatic age estimation (AAE) from face images is a challenging problem because of large facial appearance variations resulting from a number of factors, e.g., aging and facial expressions. In this paper, we propose a generic, deep ranking model for AAE. Given a face image, our network first extracts features from the face through a scattering network (ScatNet), then reduces the feature dimension by principal component analysis (PCA), and finally predicts the age via category-wise rankers. The robustness of our approach comes from the following characteristics: (1) The scattering features are invariant to translation and small deformations; (2) the rank labels encoded in the network exploit the ordering relation among labels; and (3) the category-wise rankers perform age estimation within the same group. Our network achieves superior performance on a large-scale MORPH dataset and two expression ones, Lifespan and FACES.
Event(s)
26th British Machine Vision Conference, BMVC 2015
Publisher
British Machine Vision Association
Type
conference paper