Research map / 3D vision / 3D generation
Reprojection Errors as Prompts for Efficient Scene Coordinate Regression
Ting-Ru Liu, Hsuan-Kung Yang, Jou-Min Liu, Chun-Wei Huang, Tsung-Chih Chiang, Quan Kong, Norimasa Kobori, Chun-Yi Lee
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 error-guided SAM masks and confidence refinement to ACE | ACE · Indoor6 | accuracy within 5cm/5deg (%): 35.6 -> 56.1 | Improved |
| add error-guided masks (without confidence refinement) (ablation) | ACE · Indoor6 | accuracy within 5cm/5deg (%): 35.6 -> 41.6 | Improved |
| add confidence refinement (without error-guided masks) (ablation) | ACE · Indoor6 | accuracy within 5cm/5deg (%): 35.6 -> 50.9 | Improved |
| combine error-guided masks and confidence refinement (ablation) | ACE · Cambridge Landmarks | median error (cm/deg): 25/0.4 -> 17/0.3 | Improved |
| expand low-error points into SAM masks instead of sampling points under error quantile thresholds (ablation) | EGFS · Indoor6 | accuracy within 5cm/5deg (%): 44.3 -> 56.1 | Improved |
| vary point prompt proportion tau on Cambridge (ablation) | EGFS · Cambridge Landmarks | median error (cm/deg): 17/0.3 for all tau | No clear effect |
| vary point prompt proportion tau on Indoor6 from 10 (ablation) | EGFS · Indoor6 | accuracy within 5cm/5deg (%): 56.1 -> 54.3 (tau=5) | Hurt |
| use EGFS ensemble of two models (dual) versus ACE quad ensemble | EGFS · Indoor6 | accuracy within 5cm/5deg (%): 58.6 -> 61.2 | Improved |
| add EGFS training to ACE | ACE · Cambridge Landmarks | mapping time: 5 min -> 12 min | Hurt |
Closest papers by meaning
- R-SCoRe: Revisiting Scene Coordinate Regression for Robust Large-Scale Visual Localization (CVPR 2025 · similarity 0.88)
- GLACE: Global Local Accelerated Coordinate Encoding (CVPR 2024 · similarity 0.88)
- Scene-agnostic Pose Regression for Visual Localization (CVPR 2025 · similarity 0.86)
- Map-Relative Pose Regression for Visual Re-Localization (CVPR 2024 · similarity 0.84)
- LiSA: LiDAR Localization with Semantic Awareness (CVPR 2024 · similarity 0.83)
- ConDo: Continual Domain Expansion for Absolute Pose Regression (AAAI 2025 · similarity 0.83)
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.
