Paper recorded by Signals 4 on 2026-10-13 in cs.CV. Abstract reproduced from arXiv; link to the original below.
Published 2026-10-13 on arXiv · recorded by Signals 4 on 2026-10-14
Category: cs.CV · 计算机视觉 · first seen 2026-10-14
Reconstructing complete, high-quality meshes of large-scale scenes from unposed multi-view images requires faithful geometric grounding while plausibly completing surfaces that are unobserved or only sparsely captured. We present WorldRecon, a generative multi-view reconstruction framework for recovering complete, large-scale 3D scene meshes from unposed RGB images. WorldRecon uses a coarse, noisy