Research map / 3D vision / Depth and stereo
InLoc: Indoor Visual Localization With Dense Matching and View Synthesis
Hajime Taira, Masatoshi Okutomi, Torsten Sattler, Mircea Cimpoi, Marc Pollefeys, Josef Sivic, Tomas Pajdla, Akihiko Torii
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 |
|---|---|---|---|
| Use InLoc's dense matching and view-synthesis pipeline instead of direct 2D-3D matching | Indoor visual localization · InLoc, 0.50m and 10° threshold | correctly localized queries (%): 15.8 -> 56.5 | Improved |
| Use dense CNN matching for pose estimation instead of sparse local-feature matching (ablation) | NetVLAD retrieval · InLoc | localization rate: about 15% | Improved |
| Add pose verification using virtual view synthesis (ablation) | Top-10 candidate poses · InLoc | fraction correctly localized within 1.5m: 13% or more | Improved |
| Binarize intermediate CNN features rather than retaining float descriptors (ablation) | Dense pose estimation · InLoc, 0.5m threshold | localization performance: less than 1% | Hurt |
| Use NetVLAD retrieval rather than Disloc retrieval | Indoor localization before SparsePE re-ranking · InLoc, 0.5m threshold | localization accuracy: 5.8% | Improved |
| Replace SparsePE with DensePE while retaining NetVLAD retrieval (ablation) | Indoor localization · Matterport3D (90 buildings) | median positional and angular error: 0.44, 4.70° -> 0.23, 2.53° | Improved |
| Add DensePE to NetVLAD + SparsePE (ablation) | Indoor localization · 7 Scenes | median positional and angular error | Improved |
Later papers that compared with this method
3 later accepted papers report InLoc as a baseline in their results tables; 2 of them report a better number on most of the results they share with it.
- OFVL-MS: Once for Visual Localization across Multiple Indoor Scenes (ICCV 2023 · better on 1 of 1 shared result)
- Pose Correction for Highly Accurate Visual Localization in Large-Scale Indoor Spaces (ICCV 2021 · better on 3 of 3 shared results)
- SANet: Scene Agnostic Network for Camera Localization (ICCV 2019 · better on 1 of 8 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
- Pose Correction for Highly Accurate Visual Localization in Large-Scale Indoor Spaces (ICCV 2021 · similarity 0.91)
- SPVLoc: Semantic Panoramic Viewport Matching for 6D Camera Localization in Unseen Environments (ECCV 2024 · similarity 0.90)
- 360Loc: A Dataset and Benchmark for Omnidirectional Visual Localization with Cross-device Queries (CVPR 2024 · similarity 0.89)
- Is This the Right Place? Geometric-Semantic Pose Verification for Indoor Visual Localization (ICCV 2019 · similarity 0.88)
- Rooms from Motion: Un-posed Indoor 3D Object Detection as Localization and Mapping (NeurIPS 2025 · similarity 0.86)
- RELOCATE: A Simple Training-Free Baseline for Visual Query Localization Using Region-Based Representations (CVPR 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.
