Research map / 3D vision / Depth and stereo
SPVLoc: Semantic Panoramic Viewport Matching for 6D Camera Localization in Unseen Environments
Niklas Gard, Anna Hilsmann, Peter Eisert
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 |
|---|---|---|---|
| estimate full 6D pose with panorama viewport matching versus LASER 2D floor plan localization | SPVLoc · ZInD 90° FoV | recall at 1m: 80.90 -> 90.84 | Improved |
| use SPVLoc instead of LASER for yaw rotation | SPVLoc · ZInD 90° FoV | median rotation error (deg): 0.77 -> 1.34 | Hurt |
| add refinement by re-rendering panorama at estimated pose (ablation) | SPVLoc · ZInD 90° FoV | recall at 10cm: 26.88 -> 33.69 | Improved |
| add perspective supervision and viewport decoder losses (ablation) | SPVLoc · ZInD 90° FoV | recall at 1m: 89.59 -> 91.90 | Improved |
| add negative samples from other rooms during training (ablation) | SPVLoc · ZInD 90° FoV | recall at 1m: 82.19 -> 91.90 | Improved |
| use EfficientNet-S query encoder instead of ResNet-18 (ablation) | SPVLoc · ZInD 90° FoV | median rotation error (deg): 2.56 -> 2.05 | Improved |
| replace panorama convolutions with Equiconv (ablation) | SPVLoc · ZInD 90° FoV | recall at 1m: 91.90 -> 91.70 | No clear effect |
| add normal map input to panorama encoder (ablation) | SPVLoc · ZInD 90° FoV | recall at 1m: 91.90 -> 90.84 | No clear effect |
| add depth map input to panorama encoder (ablation) | SPVLoc · ZInD 90° FoV | recall at 1m: 91.90 -> 92.15 | Improved |
| train with variable field of view 45-135° instead of fixed 90° (ablation) | SPVLoc · ZInD input FoV 80° | median translation error (cm): 37.51 -> 16.75 | Improved |
| use coarser panorama sampling grid of 2.0m instead of 1.2m (ablation) | SPVLoc · ZInD 90° FoV | recall at 1m: -9.05 | Hurt |
Later papers that compared with this method
1 later accepted paper reports SPVLoc as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- PGGT-Loc: Primitive-Grounded Geometry Transformer for Feed-Forward Camera Localization (NeurIPS 2026 · better on 3 of 3 shared results)
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- LASER — LASER: LAtent SpacE Rendering for 2D Visual Localization (CVPR 2022) · better on 7 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
- InLoc: Indoor Visual Localization With Dense Matching and View Synthesis (CVPR 2018 · similarity 0.90)
- 360Loc: A Dataset and Benchmark for Omnidirectional Visual Localization with Cross-device Queries (CVPR 2024 · similarity 0.87)
- Rooms from Motion: Un-posed Indoor 3D Object Detection as Localization and Mapping (NeurIPS 2025 · similarity 0.86)
- Fully Geometric Panoramic Localization (CVPR 2024 · similarity 0.86)
- Reloc3r: Large-Scale Training of Relative Camera Pose Regression for Generalizable, Fast, and Accurate Visual Localization (CVPR 2025 · similarity 0.86)
- Geometric Exploitation for Indoor Panoramic Semantic Segmentation (NeurIPS 2024 · 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.
