Research map / Graph learning / Graph algorithms and clustering
HARA: A Hierarchical Approach for Robust 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 |
|---|---|---|---|
| Use HARA instead of the strongest listed baseline | Rotation averaging · ELS, ND1, PDP, PIC, ROF, USQ, VNC, YKM | θ1 (deg): ELS: 2.8 -> 2.1; ND1: 2.7 -> 1.6; PDP: 3.5 -> 3.4; PIC: 4.6 -> 4.4; ROF: 2.8 -> 2.7; USQ: 6.3 -> 6.0; VNC: 6.2 -> 6.1; Y | Helped |
| Use HARA instead of the strongest listed baseline | Rotation averaging · GDM, TOL, ND2, ACP, ARQ | θ1 (deg): GDM: 37.5 -> 43.8; TOL: 3.9 -> 4.3; ND2: 1.1 -> 1.3; ACP: 0.8 -> 1.2; ARQ: 3.2 -> 3.6 | Hurt |
| Use HARA instead of the strongest listed baseline | Rotation averaging · ALM, MDR, MND, NYC, TFG, SNF | θ1 (deg): ALM: 3.7 -> 3.5; MDR: 5.2 -> 4.8; MND: 1.2 -> 1.1; NYC: 3.0 -> 2.9; TFG: 3.6 -> 3.5; SNF: 3.6 -> 3.6 | No clear effect |
| Add the number of valid feature matches to HARA (ablation) | HARA · 1DSfM real datasets | θ1 (deg): ELS: 2.1 -> 1.8; ND1: 1.6 -> 1.5; NYC: 2.9 -> 2.6; PIC: 4.1 -> 4.0; TFG: 3.5 -> 3.4; USQ: 5.9 -> 5.8; YKM: 3.0 -> 3.4; G | Mixed |
| Filter edges inconsistent with the initialization | HARA · Rotation averaging with moderate outlier ratios (< 0.3) | accuracy | Helped |
| Filter edges inconsistent with the initialization | HARA · Rotation averaging with higher outlier ratios | accuracy | Hurt |
Closest papers by meaning
- End-to-End Rotation Averaging With Multi-Source Propagation (CVPR 2021 · similarity 0.87)
- Hybrid Rotation Averaging: A Fast and Robust Rotation Averaging Approach (CVPR 2021 · similarity 0.86)
- ARCS: Accurate Rotation and Correspondence Search (CVPR 2022 · similarity 0.85)
- Rotation Averaging and Strong Duality (CVPR 2018 · similarity 0.85)
- Symmetry-Robust 3D Orientation Estimation (ICML 2025 · similarity 0.82)
- Multi-scale Consistency for Robust 3D Registration via Hierarchical Sinkhorn Tree (NeurIPS 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.