Reproduce “Segmenting Geometric Textures of 3D Meshes” 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

Weidan Xiong, Kaijie Qu, Shihao Feng, Shunqi Fan, Rongyu Xiong, Hui Huang

The authors present a method for segmenting repeated geometric textures on 3D meshes. It renders multi-view curvature maps, adapts SAM3 with LoRA, then uses DINOv2 features to match masks into 3D labels. In the reported comparison, it achieves 61.78% MUCoV and 70.54% Recall₅₀, versus SAMesh’s 32.35% and 21.33%; weak relief, hollow structures, and heavy occlusions remain failure cases.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.