Orthogonal Deep Features Decomposition for Age-Invariant Face Recognition

As facial appearance is subject to significant intra-class variations causedby the aging process over time, age-invariant face recognition (AIFR) remains amajor challenge in face recognition community. To reduce the intra-classdiscrepancy caused by the aging, in this paper we propose a novel approach(namely, Orthogonal Embedding CNNs, or OE-CNNs) to learn the age-invariant deepface features. Specifically, we decompose deep face features into twoorthogonal components to represent age-related and identity-related features.As a result, identity-related features that are robust to aging are then usedfor AIFR. Besides, for complementing the existing cross-age datasets andadvancing the research in this field, we construct a brand-new large-scaleCross-Age Face dataset (CAF). Extensive experiments conducted on the threepublic domain face aging datasets (MORPH Album 2, CACD-VS and FG-NET) haveshown the effectiveness of the proposed approach and the value of theconstructed CAF dataset on AIFR. Benchmarking our algorithm on one of the mostpopular general face recognition (GFR) dataset LFW additionally demonstratesthe comparable generalization performance on GFR.