Research map / 3D vision / Gaussian splatting
ULF-Loc: Unbiased Landmark Feature for Robust Visual Localization with 3D Gaussian Splatting
Yingdong Gu, Shaocheng Yan, Zhenjun Zhao, Yuan Kou, Jianxin Luo, Pengcheng Shi, Jiayuan Li
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 Keypoint-Consensus sampling to random sampling (ablation) | ULF-Loc with alpha-blending features · Cambridge Landmarks ablation | initial pose recall 15cm/5deg: 47.2 -> 60.2 | Improved |
| add Keypoint-Consensus sampling to random sampling (ablation) | ULF-Loc with GWFF · Cambridge Landmarks ablation | initial pose recall 15cm/5deg: 52.6 -> 63.1 | Improved |
| add Keypoint-Consensus sampling to farthest point sampling (ablation) | ULF-Loc with GWFF · Cambridge Landmarks ablation | initial pose recall 15cm/5deg: 33.5 -> 54.1 | Improved |
| add Keypoint-Consensus sampling to random sampling (ablation) | ULF-Loc with GWFF · Cambridge Landmarks ablation | final pose recall 15cm/5deg: 69.6 -> 70.9 | Improved |
| replace alpha-blending features with GWFF (ablation) | ULF-Loc with random sampling · Cambridge Landmarks ablation | initial pose recall 15cm/5deg: 47.2 -> 52.6 | Improved |
| replace alpha-blending features with GWFF (ablation) | ULF-Loc with random + K.C. sampling · Cambridge Landmarks ablation | final pose recall 15cm/5deg: 68.9 -> 70.9 | Improved |
| replace alpha-blending features with GWFF (ablation) | ULF-Loc with random sampling · Cambridge Landmarks ablation | final pose recall 50cm/5deg: 89.3 -> 90.9 | Improved |
| add Local Geometric Consistency Verification (ablation) | ULF-Loc · Cambridge Landmarks ablation | final pose recall 15cm/5deg: 70.9 -> 72.0 | Improved |
| add Local Geometric Consistency Verification (ablation) | ULF-Loc · Cambridge Landmarks ablation | final pose recall 50cm/5deg: 91.7 -> 93.7 | Improved |
| add geometric weighting to GWFF (ablation) | ULF-Loc with LGCV · Cambridge Landmarks ablation | final pose recall 15cm/5deg: 70.6 -> 72.0 | Improved |
| use landmark features on sparse landmarks instead of learning features for all Gaussians | ULF-Loc vs STDLoc · 7Scenes Heads | training time: 50 mins -> 5 mins | Improved |
| use sparse landmark features instead of per-Gaussian feature field | ULF-Loc vs STDLoc · 7Scenes Heads | GPU memory: 6566 MB -> 1086 MB | 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.
- GLACE — GLACE: Global Local Accelerated Coordinate Encoding (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
- Gaussian Splatting Feature Fields for (Privacy-Preserving) Visual Localization (CVPR 2025 · similarity 0.88)
- SEGS-SLAM: Structure-enhanced 3D Gaussian Splatting SLAM with Appearance Embedding (ICCV 2025 · similarity 0.86)
- GaussianUDF: Inferring Unsigned Distance Functions through 3D Gaussian Splatting (CVPR 2025 · similarity 0.86)
- FHGS: Feature-Homogenized Gaussian Splatting (NeurIPS 2025 · similarity 0.85)
- VA-GS: Enhancing the Geometric Representation of Gaussian Splatting via View Alignment (NeurIPS 2025 · similarity 0.85)
- FreeSplat: Generalizable 3D Gaussian Splatting Towards Free View Synthesis of Indoor Scenes (NeurIPS 2024 · similarity 0.85)
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.
