Reproduce “ViscoReg: Neural Signed Distance Functions via Viscosity Solutions” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Meenakshi Krishnan, Ramani Duraiswami
ViscoReg reconstructs 3D surfaces from point clouds without normal supervision by adding a diffusion-like term to the Eikonal loss early in training and decaying its weight to zero. On 260 ShapeNet shapes, mean squared Chamfer is 5.27×10⁻⁵ versus 1.36×10⁻⁴ for StEik with linear layers. Mean IoU is lower, 0.952 versus 0.963, so improvement is limited to surface distance, not every metric.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.