Research map / 3D vision / Gaussian splatting
ZeroSplat: Generalized Referring Segmentation in 3D Gaussian Splatting
Jiayu Ding, Meilu Song, Xiaoyi Zhang, Hongbo Jin, Yichen Jin, Xiangtian Si
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 VLM spatial anchoring (ablation) | ZeroSplat mask lifting · GR-ScanNet (generalized referring) | mIoU: 24.7 -> 38.6 | Improved |
| add 3D KNN diffusion alone (ablation) | ZeroSplat mask lifting · GR-ScanNet (generalized referring) | mIoU: 24.7 -> 26.2 | Improved |
| add 3D KNN diffusion on top of VLM (ablation) | ZeroSplat · GR-ScanNet (generalized referring) | mIoU: 38.6 -> 41.2 | Improved |
| add semantic label extraction (ablation) | ZeroSplat VLM module · GR-ScanNet (generalized referring) | mIoU: 24.3 -> 41.2 | Improved |
| add bounding box filter (ablation) | ZeroSplat VLM module · GR-ScanNet (generalized referring) | mIoU: 40.1 -> 41.2 | Improved |
| use geometric keyframe selection instead of random sampling (ablation) | ZeroSplat VLM module · GR-ScanNet (generalized referring) | mIoU: 39.6 -> 41.2 | Improved |
| use 3D KNN diffusion instead of 3D radius search (ablation) | ZeroSplat refinement · GR-ScanNet (generalized referring) | mIoU: 39.8 -> 41.2 | Improved |
| add spatial refinement (3D radius search) (ablation) | ZeroSplat · GR-ScanNet (generalized referring) | mIoU: 38.6 -> 39.8 | Improved |
| use Qwen3-VL-30B-A3B instead of Qwen3-VL-8B (ablation) | ZeroSplat · GR-LERF | mIoU: 49.4 -> 50.8 | Improved |
| use closed-source Qwen3-VL-Plus instead of open 30B model | ZeroSplat · GR-LERF | mIoU: 50.8 -> 56.7 | Improved |
| lower tau_safe to 0 (disable safety mechanism) (ablation) | ZeroSplat · GR-LERF waldo-kitchen | mIoU: 48.4 -> 29.3 | Hurt |
| lower tau_conf from 0.8 to 0.4 (ablation) | ZeroSplat · GR-LERF waldo-kitchen | mIoU: 48.4 -> 51.9 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 3 of 3 on most shared results.
- LUDVIG — LUDVIG: Learning-Free Uplifting of 2D Visual Features to Gaussian Splatting Scenes (ICCV 2025) · better on 3 of 3 shared results
- ReferSplat — ReferSplat: Referring Segmentation in 3D Gaussian Splatting (ICML 2025) · better on 2 of 2 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
- ReferSplat: Referring Segmentation in 3D Gaussian Splatting (ICML 2025 · similarity 0.94)
- SceneSplat++: A Large Dataset and Comprehensive Benchmark for Language Gaussian Splatting (NeurIPS 2025 · similarity 0.93)
- Dr. Splat: Directly Referring 3D Gaussian Splatting via Direct Language Embedding Registration (CVPR 2025 · similarity 0.92)
- Segment then Splat: Unified 3D Open-Vocabulary Segmentation via Gaussian Splatting (NeurIPS 2025 · similarity 0.91)
- Bootstraping Clustering of Gaussians for View-consistent 3D Scene Understanding (AAAI 2025 · similarity 0.91)
- FHGS: Feature-Homogenized Gaussian Splatting (NeurIPS 2025 · similarity 0.91)
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.
