Research map / Detection and segmentation / 3D detection and driving
RIO: 3D Object Instance Re-Localization in Changing Indoor Environments
Johanna Wald, Armen Avetisyan, Nassir Navab, Federico Tombari, Matthias Niessner
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 multi-scale network instead of single scale 120cm, both static (ablation) | RIO · 3RScan keypoint matching (dynamic patches, 95% recall) | F1: 74.17 -> 85.58 | Improved |
| use multi-scale network instead of single scale 120cm, both static (ablation) | RIO · 3RScan keypoint matching (dynamic patches, 95% recall) | Top-1: 3.94 -> 30.73 | Improved |
| add fine-tuning on dynamic patches (ablation) | RIO-multiscale · 3RScan keypoint matching (dynamic patches, 95% recall) | F1: 85.58 -> 94.37 | Improved |
| add fine-tuning on dynamic patches (ablation) | RIO-multiscale · 3RScan keypoint matching (dynamic patches, 95% recall) | Top-1: 30.73 -> 64.10 | Improved |
| add fine-tuning on dynamic patches (ablation) | RIO-singlescale 120cm · 3RScan keypoint matching (dynamic patches, 95% recall) | F1: 74.17 -> 78.71 | Improved |
| add larger context scale (120cm) to single 60cm scale input (ablation) | RIO-singlescale (static) · 3RScan keypoint matching (dynamic patches, 95% recall) | F1: 71.54 -> 74.17 | Improved |
| add fine-tuning on dynamic patches (ablation) | RIO-multiscale · 3RScan object instance re-localization (test set) | recall < 0.1m, 10deg (%): 9.92 -> 15.14 | Improved |
| add fine-tuning on dynamic patches (ablation) | RIO-multiscale · 3RScan object instance re-localization (test set) | recall < 0.2m, 20deg (%): 17.75 -> 23.76 | Improved |
| add fine-tuning on dynamic patches (ablation) | RIO-multiscale · 3RScan object instance re-localization (test set) | MRE [deg] at 0.1m, 10deg: 4.33 -> 4.75 | No clear effect |
| use RIO-multiscale (dynamic) instead of SHOT | hand-crafted descriptor · 3RScan object instance re-localization (test set) | recall < 0.2m, 20deg (%): 12.27 -> 23.76 | Improved |
| use RIO-multiscale (dynamic) instead of 3DMatch trained on static patches | learned descriptor · 3RScan object instance re-localization (test set) | recall < 0.2m, 20deg (%): 13.05 -> 23.76 | Improved |
| use RIO-D instead of FPFH on bed / sofa objects | descriptor · 3RScan object instance re-localization (test set) | accuracy (%) bed / sofa: 56.52 -> 47.83 | Hurt |
Closest papers by meaning
- Beyond Controlled Environments: 3D Camera Re-Localization in Changing Indoor Scenes (ECCV 2020 · similarity 0.90)
- Living Scenes: Multi-object Relocalization and Reconstruction in Changing 3D Environments (CVPR 2024 · similarity 0.87)
- Rooms from Motion: Un-posed Indoor 3D Object Detection as Localization and Mapping (NeurIPS 2025 · similarity 0.86)
- Open3DIS: Open-Vocabulary 3D Instance Segmentation with 2D Mask Guidance (CVPR 2024 · similarity 0.86)
- LOCATE 3D: Real-World Object Localization via Self-Supervised Learning in 3D (ICML 2025 · similarity 0.86)
- Change-Robust Online Topological Memory for Long-Term Relocalization and Semantic Navigation (NeurIPS 2026 · 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.
