Research map / 3D vision / 3D generation
LaMAR: Benchmarking Localization and Mapping for Augmented Reality
Paul-Edouard Sarlin, Mihai Dusmanu, Johannes L. Schönberger, Pablo Speciale, Lukas Gruber, Viktor Larsson, Ondrej Miksik, Marc Pollefeys
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 radio-signal filtering to image retrieval | Fusion retrieval + SP+SG · LaMAR single-frame | recall@(1deg,10cm) | Improved |
| use rig localization (generalized solver GP3P) instead of single cameras (ablation) | HoloLens 2 localization · LaMAR HoloLens 2 | recall@(1deg,10cm): 45.6 -> 64.2 | Improved |
| use SuperPoint+SuperGlue instead of SIFT matching | Fusion retrieval · LaMAR HoloLens 2 | recall@(1deg,10cm): 51.2 -> 64.2 | Improved |
| use SuperPoint+SuperGlue instead of SIFT matching | Fusion retrieval · LaMAR phone | recall@(1deg,10cm): 38.5 -> 52.2 | Improved |
| use Fusion (NetVLAD+APGeM) retrieval instead of NetVLAD | SP+SG · LaMAR HoloLens 2 | recall@(1deg,10cm): 59.9 -> 64.2 | Improved |
| add HD 360 NavVis images to the map (ablation) | SP+SG · LaMAR phone | recall@(1deg,10cm): 52.2 -> 60.6 | Improved |
| replace AR images with HD 360 images in the map (ablation) | SP+SG · LaMAR phone | recall@(1deg,10cm): 52.2 -> 47.4 | Hurt |
| replace AR images with HD 360 images in the map (ablation) | SP+SG · LaMAR HoloLens 2 | recall@(1deg,10cm): 64.2 -> 70.1 | Improved |
| select mapping pairs from GT overlap instead of retrieval+poses (ablation) | SP+SG · LaMAR phone | recall@(1deg,10cm): 52.2 -> 52.9 | No clear effect |
| lift keypoints to 3D with the lidar mesh instead of SfM triangulation (ablation) | SP+SG · LaMAR single-frame | recall@(1deg,10cm) | Improved |
| add sequence localization (PGO with tracking) with radios (ablation) | SP+SG sequence pipeline · LaMAR CAB phone | recall@(1deg,10cm): 54.2 -> 62.7 | Improved |
| add guided re-localization after PGO1 (ablation) | sequence pipeline · LaMAR CAB HL2 | recall@(5deg,1m): 92.4 -> 88.0 | Hurt |
Closest papers by meaning
- Visual Localization Using Imperfect 3D Models From the Internet (CVPR 2023 · similarity 0.89)
- AR-Nav Benchmark: Augmented Reality Navigation with Vision and Language (AAAI 2026 · similarity 0.89)
- CroCoDL: Cross-device Collaborative Dataset for Localization (CVPR 2025 · similarity 0.88)
- Benchmarking 6DOF Outdoor Visual Localization in Changing Conditions (CVPR 2018 · similarity 0.87)
- Efficient Depth Estimation for Unstable Stereo Camera Systems on AR Glasses (CVPR 2025 · similarity 0.84)
- 360Loc: A Dataset and Benchmark for Omnidirectional Visual Localization with Cross-device Queries (CVPR 2024 · similarity 0.84)
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.
