Research map / 3D vision / Depth and stereo
OFVL-MS: Once for Visual Localization across Multiple Indoor Scenes
Tao Xie, Kun Dai, Siyi Lu, Ke Wang, Zhiqiang Jiang, Jinghan Gao, Dedong Liu, Jie Xu, Lijun Zhao, Ruifeng 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 |
|---|---|---|---|
| train all scenes jointly with shared parameters instead of separate models per scene (ablation) | OFVL-MS34 · 7-Scenes | 5cm-5deg accuracy (%): 86.50 -> 87.37 | Improved |
| add learnable layer-adaptive sharing with task-specific parameters instead of sharing all parameters (ablation) | OFVL-MS34 · 7-Scenes | 5cm-5deg accuracy (%): 79.94 -> 87.37 | Improved |
| keep BN layers task-specific rather than score-determined (ablation) | OFVL-MS34 · 7-Scenes | 5cm-5deg accuracy (%): 86.10 -> 87.37 | Improved |
| add task-specific attention modules (ablation) | OFVL-MS34 · 7-Scenes | 5cm-5deg accuracy (%): 84.30 -> 87.37 | Improved |
| add gradient normalization algorithm (ablation) | OFVL-MS34 · 7-Scenes | 5cm-5deg accuracy (%): 84.16 -> 87.37 | Improved |
| add sparse penalty loss on sharing scores (ablation) | OFVL-MS34 · 7-Scenes | 5cm-5deg accuracy (%): 86.25 -> 87.37 | Improved |
| scale backbone from ResNet18 to ResNet50 | OFVL-MS · 7-Scenes | 5cm-5deg accuracy (%): 85.19 -> 88.72 | Improved |
| freeze shared parameters and add scene-specific parameters for new scenes (12-Scenes to 7-Scenes) | OFVL-MS50 · 7-Scenes | 5cm-5deg accuracy (%): 83.69 -> 86.75 | Improved |
| generalize from 7-Scenes to new 12-Scenes with frozen shared parameters | OFVL-MS50 · 12-Scenes | 5cm-5deg accuracy (%): 99.6 -> 98.3 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- VS-Net — VS-Net: Voting With Segmentation for Visual Localization (CVPR 2021) · better on 1 of 1 shared result
- InLoc — InLoc: Indoor Visual Localization With Dense Matching and View Synthesis (CVPR 2018) · 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
- FOCUS: Forcing In-Context Object Localization through Visual Support Constraints and Policy Optimization (ICML 2026 · similarity 0.87)
- RELOCATE: A Simple Training-Free Baseline for Visual Query Localization Using Region-Based Representations (CVPR 2025 · similarity 0.86)
- From Coarse to Fine: Robust Hierarchical Localization at Large Scale (CVPR 2019 · similarity 0.86)
- Hierarchical Scene Coordinate Classification and Regression for Visual Localization (CVPR 2020 · similarity 0.86)
- R-SCoRe: Revisiting Scene Coordinate Regression for Robust Large-Scale Visual Localization (CVPR 2025 · similarity 0.85)
- MutualVPR: A Mutual Learning Framework for Resolving Supervision Inconsistencies via Adaptive Clustering (NeurIPS 2025 · 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.
