Reproduce “PRNet: Self-Supervised Learning for Partial-to-Partial Registration” 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
Yue Wang, Justin M. Solomon
PRNet aligns partial 3D scans by repeatedly selecting salient points, matching them, estimating a rigid rotation and translation, and refining the alignment. It adapts match sharpness between passes and trains without labeled correspondences, using synthesized pose-supervised pairs. On synthetic partial ModelNet40 unseen clouds, PRNet reports mean absolute rotation error of 1.454°, versus 4.448° for DCP-v2, within this test setting.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.