Research map / 3D vision / Human pose estimation
Probabilistic Triangulation for Uncalibrated Multi-View 3D Human Pose Estimation
Boyuan Jiang, Lei Hu, Shihong Xia
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 |
|---|---|---|---|
| model camera pose as a distribution updated by Monte Carlo sampling instead of RANSAC 8-point estimate | Probabilistic Triangulation · Human3.6M camera pose estimation | E3D (mm): 97.50 -> 27.80 | Improved |
| use Probabilistic Triangulation instead of Bundle Adjustment for camera estimation | uncalibrated pose estimation · Human3.6M camera pose estimation | E3D (mm): 736.82 -> 27.80 | Improved |
| uncalibrated method vs prior uncalibrated Bartol et al. | Probabilistic Triangulation · CMU Panoptic (4 cameras) | MPJPE (mm): 25.4 -> 24.2 | Improved |
| rotate world coordinate system at test time (Test*) (ablation) | calibrated Iskakov et al. vs Probabilistic Triangulation · Human3.6M Test* | MPJPE (mm): Iskakov 20.8 -> 25.12; ours unchanged 27.8 | Hurt |
| use ground-truth 2D heatmaps instead of estimated ones (ablation) | Probabilistic Triangulation · Human3.6M | MPJPE (mm): 27.8 -> 18.5 | Improved |
| train on 4 views instead of 2 views, test on 3 views (ablation) | Probabilistic Triangulation · Human3.6M | MPJPE (mm): 42.4 -> 29.3 | Improved |
| cross-dataset generalization CMU to Human3.6M versus Bartol et al. | Probabilistic Triangulation · CMU1 train, Human3.6M test | MPJPE (mm): 33.5 -> 30.8 | Improved |
| occluded actions (Purchases, Sit) degrade 2D joints and camera distribution | Probabilistic Triangulation · Human3.6M Purch. and Sit. | MPJPE (mm): 36.5 and 40.1 vs Gordon 32.9 and 29.5 | Hurt |
Closest papers by meaning
- Generalizable Human Pose Triangulation (CVPR 2022 · similarity 0.92)
- Learnable Triangulation of Human Pose (ICCV 2019 · similarity 0.90)
- Consensus-Based Optimization for 3D Human Pose Estimation in Camera Coordinates (IJCV 2022 · similarity 0.89)
- Unified 2D-3D Discrete Priors for Noise-Robust and Calibration-Free Multiview 3D Human Pose Estimation (NeurIPS 2025 · similarity 0.88)
- ProPose: Probabilistic 3D Human Pose Estimation with Instance-Level Distribution and Normalizing Flow (AAAI 2025 · similarity 0.88)
- PoseIRM: Enhance 3D Human Pose Estimation on Unseen Camera Settings via Invariant Risk Minimization (CVPR 2024 · 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.
