Research map / 3D vision / Depth and stereo
SceneSqueezer: Learning To Compress Scene for Camera Relocalization
Luwei Yang, Rakesh Shrestha, Wenbo Li, Shuaicheng Liu, Guofeng Zhang, Zhaopeng Cui, Ping Tan
Check 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 SceneSqueezer instead of QP+RootSIFT scene compression | Cambridge Landmarks localization · Cambridge Landmarks: Shop Facade | median translation error (m): 0.72 -> 0.11 | Improved |
| Use SceneSqueezer instead of QP+RootSIFT scene compression | Cambridge Landmarks localization · Cambridge Landmarks: Old Hospital | median translation error (m): 0.9 -> 0.37 | Improved |
| Use SceneSqueezer instead of QP+RootSIFT scene compression | Cambridge Landmarks localization · Cambridge Landmarks: King’s College | median translation error (m): 1.53 -> 0.27 | Improved |
| Use SceneSqueezer instead of QP+RootSIFT scene compression | Cambridge Landmarks localization · Cambridge Landmarks: St. Mary’s Church | median translation error (m): 0.56 -> 0.15 | Improved |
| Use SceneSqueezer instead of HLoc | Aachen Day-Night localization · Aachen Day-Night: Aachen Day | localization recall @ 0.5m, 5° (%): 95.4 -> 89.7 | Hurt |
| Use SceneSqueezer instead of HLoc | Aachen Day-Night localization · Aachen Day-Night: Aachen Night | localization recall @ 0.5m, 5° (%): 93.9 -> 67.3 | Hurt |
| Use SceneSqueezer instead of HLoc | Aachen Day-Night localization · Aachen Day-Night | scene size (MB): 7828 -> 31 | Improved |
| Use SceneSqueezer instead of Cascaded descriptor compression | Aachen Day-Night localization · Aachen Day-Night: Aachen Night | localization recall @ 0.5m, 5° (%): 48.0 -> 67.3 | Improved |
| Fuse verification-frame features using appearance and geometric similarity (ablation) | Differentiable Point Selection · Cambridge Landmarks | percentage of selected points with under 10px reprojection error | Improved |
| Add a learned feature-based term to the spatial proximity kernel (ablation) | Differentiable Point Selection · King’s College | localization recall @ 0.5m, 5° | Improved |
| Use spatial proximity with learned distinctiveness instead of selecting only top distinctiveness scores (ablation) | Differentiable Point Selection · King’s College | localization recall @ 0.5m, 5° | Improved |
| Use learned point selection instead of random point selection (ablation) | Differentiable Point Selection · King’s College | localization recall @ 0.5m, 5° | Improved |
Later papers that compared with this method
1 later accepted paper reports SceneSqueezer as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- MAD-DR: Map Compression for Visual Localization with Matchness Aware Descriptor Dimension Reduction (ECCV 2024 · better on 5 of 5 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
- Simple but Effective Triplet-Based Compression Strategies for Compact Visual Localization (CVPR 2026 · similarity 0.90)
- Hybrid Scene Compression for Visual Localization (CVPR 2019 · similarity 0.90)
- Accelerated Coordinate Encoding: Learning to Relocalize in Minutes Using RGB and Poses (CVPR 2023 · similarity 0.87)
- R-SCoRe: Revisiting Scene Coordinate Regression for Robust Large-Scale Visual Localization (CVPR 2025 · similarity 0.86)
- RELOCATE: A Simple Training-Free Baseline for Visual Query Localization Using Region-Based Representations (CVPR 2025 · similarity 0.85)
- Learning to Produce Semi-dense Correspondences for Visual Localization (CVPR 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.
