Reproduce “GVKF: Gaussian Voxel Kernel Functions for Highly Efficient Surface Reconstruction in Open Scenes” 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

Gaochao Song, Chong Cheng, Hao Wang

GVKF reconstructs outdoor scenes as continuous surfaces while retaining Gaussian splat rendering. It organizes Gaussians in sparse voxels, models their influence as a continuous opacity field, and maps it to a surface. On Waymo, it reports PSNR of 30.24 versus 28.51 for 2DGS, with lower storage and memory but slower rendering and training; its Chamfer distance is worse than StreetSurf’s.

Summary by alphaXiv.

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