Research map / 3D vision / 3D generation
Learning 3D Representations for Spatial Intelligence from Unposed Multi-View Images
Bo Zhou, Qiuxia Lai, Zeren Sun, Xiangbo Shu, Yazhou Yao, Wenguan Wang
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add self-supervised appearance learning (ablation) | UniSplat · ScanNet ablation | mIoU: 0.5263 -> 0.5625 | Improved |
| add dual masking (ablation) | UniSplat · ScanNet ablation | PSNR: 24.74 -> 25.65 | Improved |
| add dual masking (ablation) | UniSplat · ScanNet ablation | mIoU: 0.5462 -> 0.5625 | Improved |
| add coarse-to-fine Gaussian splatting (ablation) | UniSplat · ScanNet ablation | mIoU: 0.5374 -> 0.5625 | Improved |
| add pose-conditioned recalibration loss (ablation) | UniSplat · ScanNet ablation | PSNR: 24.35 -> 25.65 | Improved |
| add pose-conditioned recalibration loss (ablation) | UniSplat · ScanNet ablation | Abs Rel: 3.52 -> 3.10 | Improved |
| add semantic distillation loss (ablation) | UniSplat · ScanNet ablation | mIoU: 0.0214 -> 0.5625 | Improved |
| add geometric prior loss (ablation) | UniSplat · ScanNet ablation | Abs Rel: 3.59 -> 3.10 | Improved |
| use geometry-aware decoder mask (3D GS, 0.50) instead of random-only masking (ablation) | UniSplat · ScanNet mask ablation | mIoU: 0.5412 -> 0.5625 | Improved |
| raise encoder mask ratio from 0.50 to 0.75 (ablation) | UniSplat · ScanNet mask ablation | mIoU: 0.5625 -> 0.5284 | Hurt |
| add ScanNet++ training data (ablation) | UniSplat · ScanNet data scale | mIoU: 0.5603 -> 0.5625 | Improved |
| add RealEstate10K and DL3DV training data (ablation) | UniSplat · ScanNet data scale | mIoU: 0.5625 -> 0.5755 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- pixelSplat — pixelSplat: 3D Gaussian Splats from Image Pairs for Scalable Generalizable 3D Reconstruction (CVPR 2024) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Large Spatial Model: End-to-end Unposed Images to Semantic 3D (NeurIPS 2024 · similarity 0.90)
- Imagine3D-LLM: Teaching MLLMs to Imagine 3D Scenes Before Answering (NeurIPS 2026 · similarity 0.90)
- Beyond 3D VQAs: Injecting 3D Spatial Priors into Vision-Language Models for Enhanced Geometric Reasoning (CVPR 2026 · similarity 0.89)
- UniUGG: Unified 3D Understanding and Generation via Geometric-Semantic Encoding (ICLR 2026 · similarity 0.89)
- G$^2$VLM: Geometry Grounded Vision Language Model with Unified 3D Reconstruction and Spatial Reasoning (CVPR 2026 · similarity 0.89)
- SphMind: Towards Robust, Training-Free VLM-based Spatial Reasoning with a 360 Camera (NeurIPS 2026 · similarity 0.89)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
