Research map / 3D vision / Gaussian splatting
Rh-3DGS: Robust Open-Vocabulary Scene Understanding via Riemannian Huber Distillation and Manifold-Aware Sampling
Xinpeng Zhao, Jiang Jie, Fengyuan Zhang, Lixin Zhan, Dong Wang, Qinyuan Bu, Jiahang Tu, Guangzhen Yao
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 VCD, VFM and LIC together (ablation) | 3DGS baseline · LERF (figurines) | mIoU: 60.56 -> 81.62 | Improved |
| add VCD to VFM+LIC (ablation) | Rh-3DGS · LERF (figurines) | mIoU: 76.24 -> 81.62 | Improved |
| add LIC to VCD+VFM (ablation) | Rh-3DGS · LERF (figurines) | mIoU: 75.08 -> 81.62 | Improved |
| add VCD alone (ablation) | 3DGS baseline · LERF (figurines) | mIoU: 60.56 -> 65.37 | Improved |
| add VFM pixel-level Huber geodesic loss only (ablation) | 3DGS baseline · LERF (figurines) | mIoU: 60.56 -> 71.81 | Improved |
| add Frechet-mean term only (ablation) | 3DGS baseline · LERF (figurines) | mIoU: 60.56 -> 66.46 | Improved |
| add Frechet-mean term to pixel loss (full VFM) (ablation) | VFM · LERF (figurines) | mIoU: 71.81 -> 73.76 | Improved |
| start LIC from iteration 0 instead of 3k (ablation) | Rh-3DGS · LERF (figurines) | mIoU: 81.62 -> 80.94 | Hurt |
| refresh pseudo-instances every 1000 iterations instead of 500 (ablation) | Rh-3DGS · LERF (figurines) | mIoU: 81.62 -> 81.17 | Hurt |
| add all components (inference cost) (ablation) | Rh-3DGS · LERF (figurines) rendering | FPS: 328.59 -> 301.64 | Hurt |
| replace Rh-3DGS for best baseline | Rh-3DGS vs ILGS · ScanNet 19 classes | mIoU: 36.8 -> 40.8 | 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.
- LangSplat — LangSplat: 3D Language Gaussian Splatting (CVPR 2024) · better on 2 of 2 shared results
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
- Cross-Modal and Uncertainty-Aware Agglomeration for Open-Vocabulary 3D Scene Understanding (CVPR 2025 · similarity 0.90)
- Bootstraping Clustering of Gaussians for View-consistent 3D Scene Understanding (AAAI 2025 · similarity 0.90)
- Tackling View-Dependent Semantics in 3D Language Gaussian Splatting (ICML 2025 · similarity 0.90)
- Rethinking Open-Vocabulary Segmentation of Radiance Fields in 3D Space (AAAI 2025 · similarity 0.89)
- FHGS: Feature-Homogenized Gaussian Splatting (NeurIPS 2025 · similarity 0.89)
- Language Embedded 3D Gaussians for Open-Vocabulary Scene Understanding (CVPR 2024 · similarity 0.88)
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.
