Reproduce “SAGE: Semantic-Augmented Geometric Experts for Point Cloud 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
Tomoyasu Shimada, Kazuhiko Murasaki, Shoichiro Takeda, Shogo Sato, Riku Inoue, Toshihiko Nishimura, Ryuichi Tanida
SAGE aligns partial 3D scans by combining frozen semantic features with geometric features and routing superpoints to experts using semantics alone. Its experts refine semantic-geometric features before matching correspondences and estimating alignment. With GeoTransformer’s LGR estimator, SAGE reaches 76.3% registration recall on 3DLoMatch, versus 74.0% for GeoTransformer; on 3DMatch, it trails GeoTransformer, 91.0% to 92.5%.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.