Research map / 3D vision / Human pose estimation
Affine Correspondences Between Multi-Camera Systems for Relative Pose Estimation
Banglei Guan, Ji Zhao
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 |
|---|---|---|---|
| use two ACs instead of PCs (2AC method vs 6PC-Stewenius runtime) | 6DOF multi-camera solver · Solver runtime | runtime (us): 3275.4 -> 1084.8 | Improved |
| add extra affine constraints E2 (2AC-inter-48 vs 2AC-inter-56) (ablation) | 2AC-inter solver · Solver runtime | runtime (us): 1084.8 -> 842.3 | Improved |
| add extra equations E2 to 2AC-inter (ablation) | 2AC-inter-56 · Noise-free numerical stability | round-off error | Hurt |
| use 2AC method vs 8PC-Kneip on nuScenes | 2AC method · nuScenes Part 1 | rotation error (deg): 0.156 -> 0.114 | Improved |
| use 2AC method vs 6AC-Ventura on KITTI RANSAC runtime | 2AC method · KITTI (11 sequences) | RANSAC runtime (s): 6.83 -> 4.87 | Improved |
| use 2AC method vs 6AC-Ventura on EuRoC | 2AC method · EuRoC (first sequence row) | translation direction error (deg): 2.858 -> 2.519 | Improved |
| use hallucinated PCs converted from ACs instead of AC point pairs (ablation) | PC-based solvers · Synthetic random motion | rotation and translation error | Hurt |
| reduce distribution area size s of hallucinated PCs from 10 to 1 pixels (ablation) | PC-based solvers · Synthetic random motion | rotation and translation error | Improved |
| increase support region side length of AC from 30 to 40 pixels (ablation) | AC-based solvers · Synthetic data | rotation and translation error | Improved |
| use inter-camera ACs instead of intra-camera ACs (ablation) | 2AC method · Synthetic data | translation scale error | Improved |
| use linear solvers 17PC-Li and 6AC-Ventura vs 2AC under noise-free input | 17PC-Li/6AC-Ventura · Noise-free numerical stability | round-off error | Improved |
| fewer correspondences: 2 instead of 6 at 50% outliers | RANSAC iterations · RANSAC iteration count | iterations: 439 -> 25 | Improved |
Closest papers by meaning
- Affine Correspondences between Multi-Camera Systems for 6DOF Relative Pose Estimation (ECCV 2022 · similarity 0.99)
- From Correspondences to Pose: Non-minimal Certifiably Optimal Relative Pose without Disambiguation (CVPR 2024 · similarity 0.91)
- Practical Solutions to the Relative Pose of Three Calibrated Cameras (CVPR 2025 · similarity 0.90)
- FAR: Flexible Accurate and Robust 6DoF Relative Camera Pose Estimation (CVPR 2024 · similarity 0.88)
- Relative Pose Estimation through Affine Corrections of Monocular Depth Priors (CVPR 2025 · similarity 0.88)
- Learning Affine Correspondences by Integrating Geometric Constraints (CVPR 2025 · similarity 0.87)
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.
