Convolutional Neural Networks on Surfaces via Seamless Toric Covers


Published August 31, 2017

Haggai Maron, Meirav Galun, Noam Aigerman, Miri Trope, Nadav Dym, Ersin Yumer, Vladimir Kim, Yaron Lipman

The recent success of convolutional neural networks (CNNs) for image processing tasks is inspiring research eorts attempting to achieve similar success for geometric tasks. One of the main challenges in applying CNNs to surfaces is dening a natural convolution operator on surfaces. In this paper we present a method for applying deep learning to sphere- type shapes using a global seamless parameterization to a planar at-torus, for which the convolution operator is well dened. As a result, the standard deep learning framework can be readily applied for learning semantic, high- level properties of the shape. An indication of our success in bridging the gap between images and surfaces is the fact that our algorithm succeeds in learning semantic information from an input of raw low-dimensional feature vectors. We demonstrate the usefulness of our approach by presenting two ap- plications: human body segmentation, and automatic landmark detection on anatomical surfaces. We show that our algorithm compares favorably with competing geometric deep-learning algorithms for segmentation tasks, and is able to produce meaningful correspondences on anatomical surfaces where hand-crafted features are bound to fail.

Learn More