Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes
Revisiting Rotation Averaging: Uncertainties and Robust Losses
Ganlin Zhang, Viktor Larsson, Daniel Barath
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 |
|---|---|---|---|
| Add covariance-based uncertainty weighting and use MAGSAC loss | Global SfM rotation averaging · 1DSfM, full pipeline | AUC at 5°: 59.4 -> 67.6 | Improved |
| Add covariance-based uncertainty weighting with MAGSAC loss | Global SfM rotation averaging · ETH3D, full pipeline | AUC at 2°: 86.9 -> 91.2 | Improved |
| Replace Soft L1 with MAGSAC (ablation) | Rotation averaging on 1DSfM · 1DSfM | AUC at 2°: 30.4 -> 35.9 | Improved |
| Use Tukey instead of MAGSAC (ablation) | Rotation averaging on 1DSfM · 1DSfM | AUC at 20°: 84.0 -> 84.7 | Improved |
| Replace full covariance weighting with trace weighting (ablation) | Uncertainty-aware rotation averaging · 1DSfM | AUC at 5°: 62.5 -> 58.8 | Hurt |
| Replace full covariance weighting with F-norm weighting (ablation) | Uncertainty-aware rotation averaging · 1DSfM | AUC at 5°: 62.5 -> 58.2 | Hurt |
| Weight edges by inlier number rather than covariance (ablation) | Global SfM with MAGSAC · 1DSfM, full pipeline | AUC at 5°: 67.6 -> 62.3 | Hurt |
| Use covariance weighting with Soft L1 instead of MAGSAC (ablation) | Global SfM rotation averaging · ETH3D, full pipeline | AUC at 2°: 91.2 -> 84.5 | Hurt |
| Use the proposed pipeline instead of baseline | Global SfM on downsampled ETH3D · ETH3D, downsample factor 8 | AUC at 2°: 80.69 -> 84.62 | Improved |
Closest papers by meaning
- Robust Relative Rotation Averaging (TPAMI 2017 · similarity 0.89)
- Shonan Rotation Averaging: Global Optimality by Surfing SO(p)(n) (ECCV 2020 · similarity 0.87)
- Hybrid Rotation Averaging: A Fast and Robust Rotation Averaging Approach (CVPR 2021 · similarity 0.86)
- Rotation Averaging with the Chordal Distance: Global Minimizers and Strong Duality (TPAMI 2019 · similarity 0.86)
- Certifiably Optimal Anisotropic Rotation Averaging (ICCV 2025 · similarity 0.86)
- Non-Rigid Structure-from-Motion: Temporally-Smooth Procrustean Alignment and Spatially-Variant Deformation Modeling (CVPR 2024 · similarity 0.83)
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.
