Multi-Task Pose-Invariant Face Recognition - 2015

PROJECT TITLE :

Multi-Task Pose-Invariant Face Recognition - 2015

ABSTRACT:

Face pictures captured in unconstrained environments usually 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 total vary of pose variations among ±90° of yaw. The proposed framework initial transforms the original create-invariant face recognition problem into a partial frontal face recognition problem. A sturdy patch-based mostly face illustration theme is then developed to represent the synthesized partial frontal faces. For each patch, a metamorphosis dictionary is learnt underneath the proposed multi-task learning scheme. The transformation dictionary transforms the options of different poses into a discriminative subspace. Finally, face matching is performed at patch level instead of at the holistic level. Intensive and systematic experimentation on FERET, CMU-PIE, and Multi-PIE databases shows that the proposed methodology consistently outperforms single-task-based baselines plus state-of-the-art methods for the pose drawback. We more extend the proposed algorithm for the unconstrained face verification downside and achieve prime-level performance on the difficult LFW data set.

Did you like this research project?

To get this research project Guidelines, Training and Code... Click Here