Reproduce “DiGS : Divergence guided shape implicit neural representation for unoriented point clouds” 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
Yizhak Ben-Shabat, Chamin Hewa Koneputugodage, Stephen Gould
DiGS reconstructs surfaces from unoriented point clouds without normal-estimation preprocessing. It learns a signed distance field, penalizing abrupt changes in its gradient before easing that constraint to allow detail. Across five SRB shapes, it reports mean Chamfer distance 0.19 and Hausdorff distance 3.52, versus 0.22 and 4.96 for PHASE+FF, a no-normal baseline; this comparison is not against methods using normals.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.