Saved in:
Bibliographic Details
Main Authors: Yousefi, Maryam, Bakhshandeh, Soodeh
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2512.05783
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915656814821376
author Yousefi, Maryam
Bakhshandeh, Soodeh
author_facet Yousefi, Maryam
Bakhshandeh, Soodeh
contents When depth sensors provide only 5% of needed measurements, reconstructing complete 3D scenes becomes difficult. Autonomous vehicles and robots cannot tolerate the geometric errors that sparse reconstruction introduces. We propose curvature regularization through a discrete Laplacian operator, achieving 18.1% better reconstruction accuracy than standard variational autoencoders. Our contribution challenges an implicit assumption in geometric deep learning: that combining multiple geometric constraints improves performance. A single well-designed regularization term not only matches but exceeds the effectiveness of complex multi-term formulations. The discrete Laplacian offers stable gradients and noise suppression with just 15% training overhead and zero inference cost. Code and models are available at https://github.com/Maryousefi/GeoVAE-3D.
format Preprint
id arxiv_https___arxiv_org_abs_2512_05783
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Curvature-Regularized Variational Autoencoder for 3D Scene Reconstruction from Sparse Depth
Yousefi, Maryam
Bakhshandeh, Soodeh
Computer Vision and Pattern Recognition
Machine Learning
When depth sensors provide only 5% of needed measurements, reconstructing complete 3D scenes becomes difficult. Autonomous vehicles and robots cannot tolerate the geometric errors that sparse reconstruction introduces. We propose curvature regularization through a discrete Laplacian operator, achieving 18.1% better reconstruction accuracy than standard variational autoencoders. Our contribution challenges an implicit assumption in geometric deep learning: that combining multiple geometric constraints improves performance. A single well-designed regularization term not only matches but exceeds the effectiveness of complex multi-term formulations. The discrete Laplacian offers stable gradients and noise suppression with just 15% training overhead and zero inference cost. Code and models are available at https://github.com/Maryousefi/GeoVAE-3D.
title Curvature-Regularized Variational Autoencoder for 3D Scene Reconstruction from Sparse Depth
topic Computer Vision and Pattern Recognition
Machine Learning
url https://arxiv.org/abs/2512.05783