Evangelos Kalogerakis (from University of Massachusetts at Amherst) will be visiting STREAM on Tuesday, the 5th of June, and will give a talk entitled: Deep learning architectures for 3D shape analysis, synthesis, and animation.
Abstract: The emergence of low-cost 3D acquisition devices, such as the Kinect, and the appearance of large-scale shape repositories, such as ShapeNet, are revolutionizing computer graphics, making 3D content ubiquitous. The need for algorithms that understand, intelligently process and animate 3D shapes is thus greater than ever. In this talk, I will present my latest research on deep learning architectures for 3D shape analysis, synthesis, and animation. Specifically I will describe deep architectures (ShapePFCN, SplatNet) that combine image-based and surface-based networks for 3D shape recognition and segmentation. In contrast to other deep learning approaches for 3D shape processing, the proposed architectures allow fast shape processing at high resolutions, are robust to input geometric representation artifacts, combine both image and shape datasets for training, and focus their representation power on the shape surface. Towards the end of the talk, I will also discuss recent advances on deep architectures that automate 3D shape animation, in particular facial animation, in an animator-friendly manner.