Research map / 3D vision / Human pose estimation
A Pose-Only Geometric Constraint for Multi-Camera Pose Adjustment
Shunkun Liang, Banglei Guan, Bin Li, Qifeng Yu, Yang Shang
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 |
|---|---|---|---|
| replace bundle adjustment with pose-only constraint (MCPA vs MultiCol) | multi-camera optimization · Synthetic (200 poses, 10000 points) | runtime (s): 17.94 -> 7.11 | Improved |
| replace bundle adjustment with pose-only constraint (MCPA vs MultiCol) | multi-camera optimization · Synthetic (200 poses, 10000 points) | Hessian memory (MB): 95.13 -> 4.99 | Improved |
| replace bundle adjustment with pose-only constraint (MCPA vs MultiCol) | multi-camera optimization · ETH3D Terrains | runtime (s): 56.85 -> 9.51 | Improved |
| add right-base residual as primary (MCPALR vs MCPA) (ablation) | MCPA · Synthetic (50 poses, 1000 points) | average rotation error (deg): 0.0181 -> 0.0128 | Improved |
| add right-base residual as primary (MCPALR vs MCPA) (ablation) | MCPA · Synthetic (50 poses, 1000 points) | runtime (s): 0.25 -> 0.36 | Hurt |
| use maximum-roundness base observation selection instead of random, maximum-disparity or maximum-theta (ablation) | pose-only constraint · Synthetic Forward-Linear and Forward-Curve | triangulation error | Improved |
| triangulate from two well-selected base observations instead of all observations (ablation) | 3D point reconstruction · Synthetic Forward-Linear and Forward-Curve | triangulation error | Improved |
| use covariance-weighted statistically optimal triangulation (SOT) instead of NonLin or IRMP (ablation) | triangulation · Synthetic, sigma_max = 10 | reconstruction error: improvement in precision by an order of magnitude | Improved |
| use pose-only methods under low image noise | MCPA/MCPALR vs MultiCol and BACS · Synthetic | pose error | No clear effect |
| use pose-only methods under high image noise | MCPA/MCPALR vs MultiCol and BACS · Synthetic | pose error | Improved |
| use pose-only methods on ETH3D | MCPALR vs MultiCol · ETH3D Electro | average translation error (%): 4.029 -> 0.510 | Improved |
Closest papers by meaning
- A Pose-Only Solution to Visual Reconstruction and Navigation (TPAMI 2021 · similarity 0.90)
- Stable and Scalable Bundle Adjustment of Holistic 3D Structures (ECCV 2026 · similarity 0.88)
- Affine Correspondences Between Multi-Camera Systems for Relative Pose Estimation (TPAMI 2025 · similarity 0.87)
- Sparse-view Pose Estimation and Reconstruction via Analysis by Generative Synthesis (NeurIPS 2024 · similarity 0.86)
- LEAP: Liberate Sparse-View 3D Modeling from Camera Poses (ICLR 2024 · similarity 0.86)
- DUSt3R: Geometric 3D Vision Made Easy (CVPR 2024 · similarity 0.86)
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.
