Research map / 3D vision / Human pose estimation
Rolling Shutter Relative Pose Estimation Made Practical
Daniel Barath
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 RS-aware 7-AC solver instead of GS 5-point | GS-5PC · TUM-RS (AffNet, stride 10) | pose AUC@5deg: 0.428 -> 0.502 | Improved |
| use 7-AC solver instead of RS-44PC | RS-44PC · TUM-RS (AffNet, stride 10) | pose AUC@5deg: 0.267 -> 0.502 | Improved |
| use 7-AC solver instead of RS-44PC with RoMa | RS-44PC · TUM-RS (RoMa, stride 10) | pose AUC@5deg: 0.870 -> 0.897 | Improved |
| add RS affine constraints for velocity estimation | RS-44PC · TUM-RS (AffNet, stride 10) | translational velocity error (m): 13.110 -> 0.056 | Improved |
| use 7-AC solver in global-shutter data | GS-5PC · EuRoC MAV GS (RoMa, stride 10) | pose AUC@5deg: 0.521 -> 0.512 | No clear effect |
| use RS-aware 7-AC solver instead of RS-44PC on GS data | RS-44PC · EuRoC MAV GS (RoMa, stride 10) | pose AUC@5deg: 0.499 -> 0.512 | Improved |
| use RS-aware 7-AC solver instead of RS-44PC in synthetic oracle setting | RS-44PC · Synthetic (sigma_p=0.5px, alpha=0.5) | median rotation error (deg): 0.46 -> 0.65 | Hurt |
| use RS-aware 7-AC solver instead of GS-5PC in synthetic setting | GS-5PC · Synthetic (sigma_p=0.5px, alpha=0.5) | median rotation error (deg): 4.21 -> 0.65 | Improved |
| use affine constraints for translational velocity in synthetic setting | RS-20PC/RS-44PC · Synthetic | median velocity error: ~25 -> ~10 | Improved |
| use 7-AC solver for angular velocity in synthetic oracle setting | RS-20PC/RS-44PC · Synthetic | angular velocity error: ~2x higher error | Hurt |
| reduce RANSAC sample size from 20/44 to 7 (runtime) | RS-44PC · TUM-RS (AffNet) | runtime (s): 25.2 -> 3.7 | Improved |
| vary affine residual weight w_ac in [0.5, 2] (ablation) | RS-7AC LM refinement · TUM-RS | pose AUC | No clear effect |
Closest papers by meaning
- Single-Scanline Relative Pose Estimation for Rolling Shutter Cameras (ICCV 2025 · similarity 0.92)
- Order-One Rolling Shutter Cameras (CVPR 2025 · similarity 0.91)
- Rolling Shutter Camera Self-Calibration (ECCV 2026 · similarity 0.90)
- Rolling Shutter Camera Absolute Pose (TPAMI 2019 · similarity 0.90)
- Rolling Shutter Camera Relative Pose: Generalized Epipolar Geometry (CVPR 2016 · similarity 0.89)
- Rolling Shutter Absolute Pose Problem With Known Vertical Direction (CVPR 2016 · similarity 0.89)
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.
