Neural Varifold

Neural varifolds represent a novel approach to analyzing the geometry of 3D point cloud data by encoding both point positions and their associated tangent spaces as a measure. This representation allows for robust comparison of shapes, even with varying sampling densities or parameterizations, through the computation of varifold norms, often implemented using neural networks. Current research focuses on developing efficient neural varifold algorithms for tasks like shape matching, classification, and reconstruction, demonstrating improved performance compared to traditional methods. This technique offers a powerful and flexible framework for geometric deep learning applications, particularly in fields dealing with 3D data from sources like LiDAR and depth sensors.

Papers

November 5, 2024

Self Supervised Networks for Learning Latent Space Representations of Human Body Scans and Motions
Emmanuel Hartman, Nicolas Charon, Martin Bauer
Self Supervised Motion Information Latent Space Representation Deep Motion Motion Interpolation Neural Varifold

July 5, 2024

Neural varifolds: an aggregate representation for quantifying the geometry of point clouds
Juheon Lee, Xiaohao Cai, Carola-Bibian Schönlieb, Simon Masnou
Point Cloud Geometric Analysis Aggregated Representation Neural Varifold

July 7, 2023

VariGrad: A Novel Feature Vector Architecture for Geometric Deep Learning on Unregistered Data
Emmanuel Hartman, Emery Pierson
Neural Network Architecture Geometric Deep Learning Untranscribed Data Neural Varifold

Neural Varifold

Papers

Self Supervised Networks for Learning Latent Space Representations of Human Body Scans and Motions

Neural varifolds: an aggregate representation for quantifying the geometry of point clouds

VariGrad: A Novel Feature Vector Architecture for Geometric Deep Learning on Unregistered Data