Reproduce “From Views to Worlds: Active Exploration over 3D Worlds for Vision-Language Models” 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
Qijian Tian, Jiayu Ying, Ke Fan, Lizhuang Ma, Xin Tan
V2W lets a model explore a reconstructed scene to gather visual and spatial evidence before answering questions. It combines estimated geometry and camera poses with language-grounded objects, exposed through visual and linguistic interfaces. On MindCube-Tiny, trained V2W with Qwen2.5-VL-7B scores 90.8% overall versus SpatialDreamer’s reported 84.9%; it ties on Rotation and scores lower on Around, in a non-controlled comparison.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.