Non-planar object deformations result in challenging but informative signal variations. We aim to recover this information in a feedforward manner by employing discriminatively trained convolutional networks. We formulate the task as a regression problem and train our networks by leveraging upon manually annotated correspondences between images and 3D surfaces. In this talk, the focus will be on our recent work "DensePose", where we form the "COCO-DensePose" dataset by introducing an efficient annotation pipeline to collect correspondences between 50K persons appearing in the COCO dataset and the SMPL 3D deformable human-body model. We use our dataset to train CNN-based systems that deliver dense correspondences 'in the wild', namely in the presence of background, occlusions, multiple objects and scale variations. We experiment with fully-convolutional networks and region-based DensePose-RCNN model and observe a superiority of the latter; we further improve accuracy through cascading, obtaining a system that delivers highly accurate results in real time (http://densepose.org).
Biography: Rıza Alp Güler received B.S. and M.S. degrees from Sabanci University, Electronics Engineering Program in 2012 and 2014 respectively. Since Fall 2015, he is pursuing his PhD degree at École Centrale Paris as part of "INRIA Galen Research Team" and "Center for Visual Computing" under the supervision of Iasonas Kokkinos. His main research interest is on "learning-based models for image to surface corrspondence" and he has worked on "shape representation techniques".