Deeply-learned hybrid representations for facial age estimation

Abstract

In this paper, we propose a novel unified network named Deep Hybrid-Aligned Architecture for facial age estimation. It contains global, local and global-local branches, which are jointly optimized and thus can capture multiple types of features with complementary information. In each sub-network of each branch, we employ a separate loss to extract the independent region features and use a recurrent fusion to explore correlations among them. Considering that pose variations may lead to misalignment in different regions, we design an Aligned Region Pooling operation to generate aligned region features. Moreover, a new large private age dataset named Web-FaceAge owning more than 120K samples is collected under diverse scenes and spanning a large age range. Experiments on five age benchmark datasets, including Web-FaceAge, Morph, FG-NET, CACD and Chalearn LAP 2015, show that the proposed method outperforms the state-of-the-art approaches significantly.