PROJECT TITLE:

Multi-task Pose-Invariant Face Recognition - 2015

ABSTRACT:

Face pictures captured in unconstrained environments typically contain significant create variation, which dramatically degrades the performance of algorithms designed to acknowledge frontal faces. This paper proposes a completely unique face identification framework capable of handling the complete vary of pose variations inside ±90° of yaw. The proposed framework first transforms the first pose-invariant face recognition downside into a partial frontal face recognition drawback. A robust patch-primarily based face illustration scheme is then developed to represent the synthesized partial frontal faces. For each patch, a change dictionary is learnt beneath the proposed Multi-Task Learning scheme. The transformation dictionary transforms the features of different poses into a discriminative subspace. Finally, face matching is performed at patch level rather than at the holistic level. Extensive and systematic experimentation on FERET, CMU-PIE, and Multi-PIE databases shows that the proposed method consistently outperforms single-task-based baselines also state-of-the-art ways for the cause drawback. We any extend the proposed algorithm for the unconstrained face verification problem and achieve high-level performance on the challenging LFW data set.