Research map / Point clouds / Point cloud classification
Simple but Effective Triplet-Based Compression Strategies for Compact Visual Localization
Torsten Sattler, Zuzana Kukelova
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 PROSAC sampling instead of uniform for triplets (ablation) | TTS · Cambridge Landmarks (all scenes) | localization recall: slightly better at low memory; large drop on King's College | Mixed |
| use combined sampling instead of uniform (ablation) | TTS · Cambridge Landmarks | localization recall | No clear effect |
| vary threshold tau (ablation) | TTS · Cambridge Landmarks | localization recall | No clear effect |
| use TTS vs Greedy Set Cover, Track Length, QP baselines | structure compression baselines · Cambridge Landmarks and 7Scenes | memory-accuracy tradeoff | Improved |
| use TTS vs Track Length on small models | Track Length · Aachen Night | memory-accuracy tradeoff: TTS best above 5MB; Track Length slightly better for smaller | Mixed |
| use TTS instead of extracting fewer features (Max. Feats.) | Max. Feats. · Aachen Day-Night v1.0 | mean recall | Improved |
| apply TTS on top of at most 512 features per image (ablation) | Max. Feats. · Aachen (aliked) | mean recall | Improved |
| ignore radial distortion (ablation) | TTS · Cambridge King's College | recall: 58.0 -> 23.6 | Hurt |
| add FPS to TTS (ablation) | TTS (N=6) · 7Scenes | recall and memory: 73.0 -> 71.3; 2.82 -> 1.27 MB | Mixed |
Closest papers by meaning
- Hybrid Scene Compression for Visual Localization (CVPR 2019 · similarity 0.92)
- SceneSqueezer: Learning To Compress Scene for Camera Relocalization (CVPR 2022 · similarity 0.90)
- MAD-DR: Map Compression for Visual Localization with Matchness Aware Descriptor Dimension Reduction (ECCV 2024 · similarity 0.89)
- Differentiable Product Quantization for Memory Efficient Camera Relocalization (ECCV 2024 · similarity 0.88)
- M3SOT: Multi-Frame, Multi-Field, Multi-Space 3D Single Object Tracking (AAAI 2024 · similarity 0.86)
- 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.
