10.1007/978-3-030-17798-0_36 ; arXiv:1808.04848">
 

Ursa: A Neural Network for Unordered Point Clouds Using Constellations

Document Type

Conference Proceeding

Publication Date

2020

Abstract

This paper describes a neural network layer, named Ursa, that uses a constellation of points to learn classification information from point cloud data. Unlike other machine learning classification problems where the task is to classify an individual high-dimensional observation, in a point-cloud classification problem the goal is to classify a set of d-dimensional observations. Because a point cloud is a set, there is no ordering to the collection of points in a point-cloud classification problem. Thus, the challenge of classifying point clouds inputs is in building a classifier which is agnostic to the ordering of the observations, yet preserves the d-dimensional information of each point in the set. This research presents Ursa, a new layer type for an artificial neural network which achieves these two properties. Similar to new methods for this task, this architecture works directly on d-dimensional points rather than first converting the points to a d-dimensional volume. The Ursa layer is followed by a series of dense layers to classify 2D and 3D objects from point clouds. Experiments on ModelNet40 and MNIST data show classification results comparable with current methods, while reducing the training parameters by over 50 percent.

Comments

© Springer Nature Switzerland AG 2020.
The "Link to Full Text" on this page directs to the arXiv e-print version of the conference paper, hosted at the arXiv.org repository.

This paper was published in volume 944 of the Springer book series Advances in Intelligent Systems and Computing, as cited below, and is accessible to subscribers via the DOI link on this page.

DOI

Version of record:10.1007/978-3-030-17798-0_36 ; arXiv:1808.04848

Source Publication

Computer Vision Conference, CVC 2019

Link to Full Text

Share

COinS