Geometric Attention for Prediction of Differential Properties in 3D Point Clouds

Albert Matveev1Alexey Artemov1Denis Zorin2, 1Evgeny Burnaev1

1Skolkovo Institute of Science and Technology2New York University

The 9th IAPR TC3 Workshop on Artificial Neural Networks in Pattern Recognition 2020

We demonstrate the learned point neighborhood for the case of feature detection. The color-coding on the right image indicates the relative distances of all points from the query point. Note that the bright region border does not extend to the set of points marked as sharp, meaning that the kNN would only select points from the top plane.


Estimation of differential geometric quantities in discrete 3D data representations is one of the crucial steps in the geometry processing pipeline. Specifically, estimating normals and sharp feature lines from raw point cloud helps improve meshing quality and allows us to use more precise surface reconstruction techniques. When designing a learnable approach to such problems, the main difficulty is selecting neighborhoods in a point cloud and incorporating geometric relations between the points. In this study, we present a geometric attention mechanism that can provide such properties in a learnable fashion. We establish the usefulness of the proposed technique with several experiments on the prediction of normal vectors and the extraction of feature lines.



Copy bibtex


If you have any questions about this work, please contact us under