Research map / 3D vision / Human pose estimation
Deterministic Object Pose Confidence Region Estimation
Jinghao Wang, Zhang Li, Zi Wang, Banglei Guan, Yang Shang, Qifeng Yu
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 |
|---|---|---|---|
| regress Gaussian keypoint distributions and propagate through IFT instead of heatmap keypoints plus random sampling | pose confidence region estimation · LMO | translation region volume: 109.9 -> 8.6 | Improved |
| replace sampling with IFT deterministic propagation | pose confidence region estimation · SPEED | rotation region volume: 210.6 -> 0.2 | Improved |
| replace sampling-based with deterministic propagation | pose confidence region estimation · LMO | translation coverage rate: 96.61 -> 83.24 | Hurt |
| replace sampling-based with deterministic propagation | pose confidence region estimation · SPEED | translation coverage rate: 6.40 -> 86.69 | Improved |
| use regression-based keypoint detection instead of heatmap-based | keypoint detection · SPEED | keypoint reprojection accuracy: 57.80 -> 97.09 | Improved |
| use regression-based keypoint detection instead of heatmap-based | keypoint detection · LMO | keypoint reprojection accuracy: 67.33 -> 74.45 | Improved |
| use regression-based keypoint detection instead of heatmaps | keypoint region time · LMO | time (s): 0.0076 -> 0.0038 | Improved |
| use Yang et al. heatmap keypoints with deterministic propagation versus our regression keypoints (ablation) | LMO · LMO, epsilon=0.1 | rotation region volume: 100.1 -> 91.3 | No clear effect |
| use heatmap keypoints on occlusion-free SPEED (ablation) | Yang et al. heatmap + deterministic propagation · SPEED | translation region volume: 0.7 -> 73.4 | Hurt |
| apply ICP calibration to our regressed Gaussians | keypoint confidence region · LMO and SPEED, epsilon=0.1 | 2D keypoint coverage rate: 90.65 vs 91.16 (LMO); 89.66 vs 88.88 (SPEED) | No clear effect |
Closest papers by meaning
- Object Pose Estimation With Statistical Guarantees: Conformal Keypoint Detection and Geometric Uncertainty Propagation (CVPR 2023 · similarity 0.91)
- UA-Pose: Uncertainty-Aware 6D Object Pose Estimation and Online Object Completion with Partial References (CVPR 2025 · similarity 0.88)
- FAR: Flexible Accurate and Robust 6DoF Relative Camera Pose Estimation (CVPR 2024 · similarity 0.86)
- Category-Level Articulated Object Pose Estimation via Pose–Shape Hypothesis Generation and Verification (ECCV 2026 · similarity 0.86)
- 3D-Aware Hypothesis & Verification for Generalizable Relative Object Pose Estimation (ICLR 2024 · similarity 0.86)
- On the Calibration of Human Pose Estimation (ICML 2024 · similarity 0.85)
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.
