Research map / 3D vision / Depth and stereo
Beyond Controlled Environments: 3D Camera Re-Localization in Changing Indoor Scenes
Johanna Wald, Torsten Sattler, Stuart Golodetz, Tommaso Cavallari, Federico Tombari
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 |
|---|---|---|---|
| test on hard-change frames instead of easy-change frames | D2-Net · RIO10 test | E_f(0.15): 0.735 -> 0.244 | Hurt |
| test on hard-change frames instead of easy-change frames | Grove v2 · RIO10 test | E_f(0.15): 0.714 -> 0.112 | Hurt |
| test on texture-less instead of well-textured frames | Active Search · RIO10 test | E_f(0.15): 0.388 -> 0.156 | Hurt |
| test on novel poses instead of not-novel poses | Grove v2 · RIO10 test | E_f(0.15): 0.715 -> 0.413 | Hurt |
| test on low-context frames instead of medium-context frames | D2-Net · RIO10 test | E_f(0.15): 0.559 -> 0.406 | Hurt |
| test on high-context frames instead of medium-context frames | HF-Net · RIO10 test | E_f(0.15): 0.132 -> 0.129 | No clear effect |
| train HF-Net on RIO10 training images | HF-Net · RIO10 test | E_f(0.15): 0.098 -> 0.288 | Improved |
| add 20-NN interpolation to retrieval | NetVLAD · RIO10 test | E_f(0.15): 0.1250 -> 0.0999 | Hurt |
| add 20-NN interpolation to retrieval | DenseVLAD · RIO10 test | E_f(0.15): 0.1240 -> 0.0967 | Hurt |
| use frame sequences of 10, 30 or 100 frames instead of single images | Active Search, D2-Net, Grove, Grove v2 · RIO10 test | DCRE | Improved |
| increase scene change (visual, geometric, semantic) | best re-localizers · RIO10 test | E_f(0.15) | Hurt |
Closest papers by meaning
- RIO: 3D Object Instance Re-Localization in Changing Indoor Environments (ICCV 2019 · similarity 0.90)
- On the Limits of Pseudo Ground Truth in Visual Camera Re-Localisation (ICCV 2021 · similarity 0.89)
- Living Scenes: Multi-object Relocalization and Reconstruction in Changing 3D Environments (CVPR 2024 · similarity 0.86)
- Rooms from Motion: Un-posed Indoor 3D Object Detection as Localization and Mapping (NeurIPS 2025 · similarity 0.86)
- Change-Robust Online Topological Memory for Long-Term Relocalization and Semantic Navigation (NeurIPS 2026 · similarity 0.86)
- HyperPose: Hypernetwork-Infused Camera Pose Localization and an Extended Cambridge Landmarks Dataset (CVPR 2025 · 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.
