Research map / 3D vision / Human pose estimation
Consensus-Based Optimization for 3D Human Pose Estimation in Camera Coordinates
Diogo Carbonera Luvizon, David Picard, Hedi Tabia
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 |
|---|---|---|---|
| add multi-view fusion with ground-truth calibration | monocular absolute pose network · Human3.6M test set | MPJPE average (mm): 52 -> 39 | Improved |
| add multi-view consensus optimization with estimated calibration | monocular absolute pose network · Human3.6M test set | MPJPE average (mm): 52 -> 45 | Improved |
| use ground-truth instead of estimated calibration for multi-view (ablation) | consensus optimization · Human3.6M validation | MPJPE (mm): 44.7 -> 36.9 | Improved |
| add a third camera view to two views (GT calibration) (ablation) | multi-view pose estimation · Human3.6M validation | MPJPE (mm): 45.7 -> 41.8 | Improved |
| add a fourth camera view to three views (GT calibration) (ablation) | multi-view pose estimation · Human3.6M validation | MPJPE (mm): 41.8 -> 36.9 | Improved |
| add the refinement network (ResNet-U) instead of ResNet cut at block 5 (ablation) | pose backbone · Human3.6M validation | MPJPE (mm): 53.7 -> 50.5 | Improved |
| use backbone cut at block 5 instead of block 4 (ablation) | ResNet backbone · Human3.6M validation | MPJPE (mm): 62.2 -> 53.7 | Improved |
| add CNN visual features to bounding box features for absolute depth (ablation) | absolute root depth estimation · Human3.6M validation | MRPE (mm): 375.4 -> 80.1 | Improved |
| add bounding box position to CNN features for absolute depth (ablation) | absolute root depth estimation · Human3.6M validation | MRPE (mm): 100.1 -> 80.1 | Improved |
| use monocular-trained network on another dataset with different intrinsics (zero-shot, with torso rescaling) | absolute pose model · KTH | PCP: 0.812 and .929 legs; 0.620 and 0.804 arms | Mixed |
| use method against best prior approach on total metrics | monocular model · MPI-INF-3DHP | MPJPE total (mm): 112.1 vs 105.2 | Mixed |
Closest papers by meaning
- Learning Monocular 3D Human Pose Estimation From Multi-View Images (CVPR 2018 · similarity 0.90)
- A Simple yet Effective Baseline for 3D Human Pose Estimation (ICCV 2017 · similarity 0.90)
- Dual Networks Based 3D Multi-Person Pose Estimation From Monocular Video (TPAMI 2022 · similarity 0.90)
- Sparseness Meets Deepness: 3D Human Pose Estimation From Monocular Video (CVPR 2016 · similarity 0.90)
- PoseIRM: Enhance 3D Human Pose Estimation on Unseen Camera Settings via Invariant Risk Minimization (CVPR 2024 · similarity 0.89)
- MV-SSM: Multi-View State Space Modeling for 3D Human Pose Estimation (CVPR 2025 · 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.
