Graph neural network improves shape and keypoint matching tasks
RBF-GNN: Rational Basis Functions for Pseudo-Coordinate based Graph Convolutions
Computer Vision and Pattern Recognition
Summary
Graphs represent data connected by links, and special neural networks help understand them better. This paper presents RBF-GNN, a graph neural network that uses smarter math functions to handle spatial information more efficiently than previous methods. The authors tested their approach on tasks like matching important points in shapes and event-based camera images and found it performs better. Their method also trains more effectively thanks to new initialization and scaling techniques.
What this means in practice
- •For computer vision teams: Improve keypoint and shape matching performance for 3D object recognition using enhanced spatial graph convolutions.
- •For robotics developers: Enhance event-based camera processing for better real-time spatial understanding in robotics applications.
Authors
Paweł Batorski, Abtin Pourhadi, Paul Swoboda
Abstract
We propose RBF-GNN, a new pseudo-coordinate based graph neural network architecture that takes into account Euclidean, spherical or angular coordinates and uses them to induce a powerful spatial inductive bias. Similar in architecture to SplineCNN, we improve upon the latter by replacing the less efficient sparse-activation based B-splines whose number grows exponentially with dimension by rational Padé basis functions. For effective training we propose a spline-subspace initialization and a variance-preserving weight rescaling. Experimentally, we evaluate on a number of popular neural network architectures that use SplineCNNs. We replace only the SplineCNNs with RBF-GNN. We achieve improved results, including on semantic keypoint matching, shape matching, event based camera computer vision tasks. We will make our implementation publicly available upon acceptance of the paper.