Gao et al.
Generative framework converts sparse panoramas into annotated multi-room 3D scenes
Holi-Spatial 360 uses correspondence-guided optimization and generative priors to reconstruct over 1,000 multi-room homes from online real-estate panoramas, producing a dataset that boosts spatial reasoning in vision-language models.
Chinese Tech
Yuanyuan Gao · Zhirui Liu · Yuning Gong · Yifei Liu · Yifan Zhan · Kuixuan Jiao · +6 more
The Hong Kong University of Science and Technology · ShanghaiTech University · Tencent · Shanghai Artificial Intelligence Laboratory · The University of Tokyo
Research Digest··3 min read
The authors present Holi-Spatial 360, a pipeline that transforms sparse 360-degree panoramas from real-estate listings into photorealistic, semantically annotated 3D environments.
Why this paper
From Tencent and 6 others
In one line
Sparse real-estate panoramas can be turned into annotated multi-room 3D Gaussian Splatting scenes that improve VLM spatial reasoning.
What we could check
- ·No code link found
- ·No weights link found
- ·No dataset link found
- ·No compute details found
- ·No stated limitations found
- ·No benchmark numbers found
Observed from the paper text and links we have. Absence here means we did not find it, not that it does not exist.
§