Research map / Optimization / Optimizers
Baseline Desensitizing in Translation Averaging
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 an angle-based bilinear objective with per-edge normalization variables instead of magnitude-based residuals | Translation averaging · Synthetic view graphs with disparate baselines | NRMSE | Helped |
| Use rotation-assisted IRLS weighting rather than naïve IRLS weighting (ablation) | BATA with random initialization · 1DSfM real-data scenes | median distance error (m): 4.3 -> 1.5 | Helped |
| Use rotation-assisted IRLS weighting rather than naïve IRLS weighting (ablation) | BATA · 1DSfM real-data scenes | outer iterations to convergence: 100 -> 57 | Helped |
| Initialize BATA from RevisedLUD rather than random initialization (ablation) | BATA without rotation-assisted weighting · Piazza Pop. (PDP) | median distance error (m): 1.7 -> 1.4 | Hurt |
| Use rotation-assisted weighting rather than no rotation weighting after convex initialization (ablation) | BATA · NYC Library (NYC) | median distance error (m): 0.7 -> 0.6 | Helped |
| Use rotation-assisted weighting rather than no rotation weighting after convex initialization (ablation) | BATA · Roman For. (ROF) | median distance error (m): 1.6 -> 1.6 | Mixed |
| Use BATA instead of LUD or Shapefit/kick | Translation averaging · 1DSfM real-data scenes | median distance error (m) | Mixed |
| Replace LUD's lower-bound scale constraint with a global scale-fixing constraint | LUD formulation · Camera location configuration under disparate baselines | shape squashing: 0.727 vs 0.747 | Helped |
| Use a lower-bound scale constraint instead of a global scale-fixing constraint | LUD formulation · Synthetic camera configurations with disparate baselines | shape preservation | Hurt |
| Use BATA rather than 1DSfM | Translation averaging · Ill-conditioned synthetic view graphs with lower connectivity and higher outlier ratio | NRMSE | Helped |
| Use BATA rather than magnitude-based methods | Translation averaging · Synthetic two-cluster data as baseline separation increases | NRMSE | Helped |
| Use BATA rather than magnitude-based methods | Translation averaging · NYC Library (NYC) sparse-camera positions | camera localization error | Helped |
Closest papers by meaning
- Sensitivity in Translation Averaging (NeurIPS 2023 · similarity 0.90)
- Revisiting Global Translation Estimation with Feature Tracks (CVPR 2024 · similarity 0.85)
- Hybrid Rotation Averaging: A Fast and Robust Rotation Averaging Approach (CVPR 2021 · similarity 0.85)
- Maximum Spatial Perturbation Consistency for Unpaired Image-to-Image Translation (CVPR 2022 · similarity 0.84)
- FAR: Flexible Accurate and Robust 6DoF Relative Camera Pose Estimation (CVPR 2024 · similarity 0.83)
- Learning to Filter Outlier Edges in Global SfM (CVPR 2025 · similarity 0.82)
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.