Research map / Graph learning / Graph algorithms and clustering
RAGO: Recurrent Graph Optimizer for Multiple Rotation Averaging
Check 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 |
|---|---|---|---|
| add recurrent GRU history to the graph updater (ablation) | RAGO · Synthetic test set | mean angular error: 0.46 -> 0.24 | Helped |
| use angular degree distance instead of L1 for the cost graph (ablation) | RAGO · Synthetic test set | mean angular error: 0.24 -> 0.28 | No clear effect |
| replace cost-graph values with null vectors (ablation) | RAGO · Synthetic test set | mean angular error: 0.24 -> 0.65 | Hurt |
| use whole-graph MRA cost rather than node-wise SRA cost (ablation) | RAGO · Synthetic test set | mean angular error: 0.24 -> 3.80 | Hurt |
| use 2 Edge Convolution layers instead of 3 (ablation) | RAGO cost-feature MPNN · Synthetic test set | mean angular error: 0.24 -> 0.41 | No clear effect |
| use 1 Edge Convolution layer instead of 3 (ablation) | RAGO cost-feature MPNN · Synthetic test set | mean angular error: 0.24 -> 0.71 | Hurt |
| initialize with random SPT instead of random rotations (ablation) | RAGO · Synthetic test set | mean angular error: 0.24 -> 0.27 | Helped |
| initialize with CleanNet-SPT instead of random rotations (ablation) | RAGO · Synthetic test set | mean angular error: 0.24 -> 0.23 | Helped |
| use RAGO rather than NeuRoRA | MRA optimization · Synthetic test set, opt. | mean angular error: 1.35 -> 0.24 | Helped |
| use RAGO rather than MSP | MRA optimization · Alamo | mean angular error: 2.89 -> 2.82 | Helped |
| use RAGO rather than NeuRoRA | MRA optimization · Gendarmenmarkt | mean angular error: 4.51 -> 5.24 | Hurt |
| use RAGO rather than MSP | MRA optimization · colosseum_exterior | mean angular error: 2.7 -> 1.97 | Helped |
Closest papers by meaning
- End-to-End Rotation Averaging With Multi-Source Propagation (CVPR 2021 · similarity 0.85)
- HARA: A Hierarchical Approach for Robust Rotation Averaging (CVPR 2022 · similarity 0.84)
- Hybrid Rotation Averaging: A Fast and Robust Rotation Averaging Approach (CVPR 2021 · similarity 0.84)
- Baseline Desensitizing in Translation Averaging (CVPR 2018 · similarity 0.83)
- From Variance to Veracity: Unbundling and Mitigating Gradient Variance in Differentiable Bundle Adjustment Layers (CVPR 2024 · similarity 0.81)
- Adaptive VIO: Deep Visual-Inertial Odometry with Online Continual Learning (CVPR 2024 · similarity 0.81)
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.