Research map / 3D vision / Human pose estimation
Deep Directional Statistics: Pose Estimation with Uncertainty Quantification
Sergey Prokudin, Peter Gehler, Sebastian Nowozin
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 |
|---|---|---|---|
| learn concentration kappa with von Mises likelihood instead of fixed kappa (ablation) | Biternion network · PASCAL3D+ | log-likelihood: -0.07 -> 1.17 | Improved |
| learn concentration kappa with von Mises likelihood instead of fixed kappa (ablation) | Biternion network · PASCAL3D+ | Acc pi/6: 0.81 -> 0.84 | Improved |
| learn concentration kappa with von Mises likelihood instead of fixed kappa (ablation) | Biternion network · PASCAL3D+ | MedErr: 12.6 -> 10.7 | Improved |
| use mixture-sCVAE instead of single von Mises | single von Mises · PASCAL3D+ | log-likelihood: 1.17 -> 1.15 | No clear effect |
| use mixture-sCVAE instead of single von Mises | single von Mises · PASCAL3D+ | MedErr: 10.7 -> 12.2 | Hurt |
| use finite mixture instead of single von Mises | single von Mises · CAVIAR-o | log-likelihood: 0.700 -> 1.87 | Improved |
| use finite mixture instead of single von Mises | single von Mises · CAVIAR-o | MAAD: 5.53 -> 4.21 | Improved |
| use finite mixture instead of single von Mises | single von Mises · TownCentre | log-likelihood: -0.57 -> -0.50 | Improved |
| use finite mixture instead of single von Mises | single von Mises · TownCentre | MAAD: 22.9 -> 23.5 | No clear effect |
| use single von Mises with learned kappa instead of fixed kappa (ablation) | Beyer et al. · IDIAP head pose (pan) | log-likelihood: 0.37 -> 0.56 | Improved |
| use single von Mises with learned kappa instead of fixed kappa | Beyer et al. · IDIAP head pose (pan) | MAAD: 5.8 -> 6.3 | Hurt |
| use mixture-CVAE instead of single von Mises | single von Mises · IDIAP head pose (tilt) | MAAD: 2.3 -> 2.9 | Hurt |
Closest papers by meaning
- Implicit-PDF: Non-Parametric Representation of Probability Distributions on the Rotation Manifold (ICML 2021 · similarity 0.88)
- Plausible Uncertainties for Human Pose Regression (ICCV 2023 · similarity 0.87)
- Camera Pose Matters: Improving Depth Prediction by Mitigating Pose Distribution Bias (CVPR 2021 · similarity 0.87)
- ProPose: Probabilistic 3D Human Pose Estimation with Instance-Level Distribution and Normalizing Flow (AAAI 2025 · similarity 0.86)
- CUPS: Improving Human Pose-Shape Estimators with Conformalized Deep Uncertainty (ICML 2025 · similarity 0.86)
- Can Generative Video Models Help Pose Estimation? (CVPR 2025 · 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.
