Research map / 3D vision / 3D generation
LASER: LAtent SpacE Rendering for 2D Visual Localization
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 LASER with semantic map labels instead of MCL | Panorama localization · ZInD | 1m recall (%): 90.15 -> 97.12 | Helped |
| Use LASER with semantic map labels instead of LaLaLoc | Panorama localization · Structured3D (Full furnishing) | 1m recall (%): 87.51 -> 95.52 | Helped |
| Use LASER with semantic map labels instead of MCL | Perspective 90° FoV localization · ZInD | 1m recall (%): 34.72 -> 59.40 | Helped |
| Use LASER with semantic map labels instead of MCL | Perspective 90° FoV localization · Structured3D (Full furnishing) | 1m recall (%): 26.81 -> 56.28 | Helped |
| Use latent-space rendering with codebooks instead of per-sample rendering and encoding | LASER · S3D sampling | Sampling Fps: 2471 ± 1328 -> 13238 ± 1890 | Helped |
| Add the refinement branch (ablation) | LASER base · ZInD panorama | median rotation error (deg): 5.45 -> 0.47 | Helped |
| Add the refinement branch (ablation) | LASER base · ZInD perspective 90° FoV | median rotation error (deg): 5.46 -> 1.15 | Helped |
| Add context loss (ablation) | LASER · ZInD perspective 90° FoV | 1m recall (%): 50.52 -> 54.55 | Helped |
| Add PointNet map encoding instead of shared codebooks per semantic label (ablation) | LASER · ZInD perspective 90° FoV | 1m recall (%): 50.26 -> 54.55 | Helped |
| Add dynamic rendering codebooks instead of fixed features per map point (ablation) | LASER · ZInD perspective 90° FoV | 1m recall (%): 10.34 -> 54.55 | Helped |
| Add circular feature structure instead of a rotation-agnostic feature (V=1) (ablation) | LASER · ZInD panorama | 1m recall (%): 58.24 -> 97.06 | Helped |
| Add circular feature structure instead of a rotation-agnostic feature (V=1) (ablation) | LASER · ZInD perspective 90° FoV | 1m recall (%): 20.11 -> 54.55 | Helped |
Closest papers by meaning
- GPVK-VL: Geometry-Preserving Virtual Keyframes for Visual Localization under Large Viewpoint Changes (CVPR 2025 · similarity 0.86)
- F3Loc: Fusion and Filtering for Floorplan Localization (CVPR 2024 · similarity 0.85)
- Rooms from Motion: Un-posed Indoor 3D Object Detection as Localization and Mapping (NeurIPS 2025 · similarity 0.85)
- 360Loc: A Dataset and Benchmark for Omnidirectional Visual Localization with Cross-device Queries (CVPR 2024 · similarity 0.84)
- PLGSLAM: Progressive Neural Scene Represenation with Local to Global Bundle Adjustment (CVPR 2024 · similarity 0.84)
- RELOCATE: A Simple Training-Free Baseline for Visual Query Localization Using Region-Based Representations (CVPR 2025 · 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.