Myface: Unconstrained face recognition

Abstract

Face verification in unconstrained images, remains a challenging problem. Many works have been proposed to solve this problem. However, the performance gap existing between the human visual system and machines in face recognition remain important. This paper makes two contributions: firstly, for improving face recognition in the wild, at least in terms of pose variations, we propose a method for aligning faces by employing single-3D face model as reference produced by FaceGen Modeller. Secondly, we developed a novel face representation technique based on Gabor Filters. The proposed approach relies on combination of Gabor magnitude and Gabor phase informations into an unified framework, which capable to surpass standard representations in the well-known FERET dataset.