Face analysis techniques have become a crucial component of human-machineinteraction in the fields of assistive and humanoid robotics. However, thevariations in head-pose that arise naturally in these environments are still agreat challenge. In this paper, we present a real-time capable 3D facemodelling framework for 2D in-the-wild images that is applicable for robotics.The fitting of the 3D Morphable Model is based exclusively on automaticallydetected landmarks. After fitting, the face can be corrected in pose andtransformed back to a frontal 2D representation that is more suitable for facerecognition. We conduct face recognition experiments with non-frontal imagesfrom the MUCT database and uncontrolled, in the wild images from the PaSCdatabase, the most challenging face recognition database to date, showing animproved performance. Finally, we present our SCITOS G5 robot system, whichincorporates our framework as a means of image pre-processing for faceanalysis.
展开▼
