ProV Logo
0

Geodesic convolutional neural networks o...
Masci, Jonathan...
Geodesic convolutional neural networks on Riemannian manifolds by Masci, Jonathan ( Author )
Australian National University
07-09-2023
Feature descriptors play a crucial role in a wide range of geometry analysis and processing applications, including shape correspondence, retrieval, and segmentation. In this paper, we introduce Geodesic Convolutional Neural Networks (GCNN), a generalization of the convolutional networks (CNN) paradigm to non-Euclidean manifolds. Our construction is based on a local geodesic system of polar coordinates to extract "patches", which are then passed through a cascade of filters and linear and non-linear operators. The coefficients of the filters and linear combination weights are optimization variables that are learned to minimize a task-specific cost function. We use GCNN to learn invariant shape features, allowing to achieve state-of-the-art performance in problems such as shape description, retrieval, and correspondence.
-
Article
pdf
29.34 KB
English
-
MYR 0.01
-
http://arxiv.org/abs/1501.06297
Share this eBook