Research map / 3D vision / Human pose estimation
Monocular 3D Human Pose Estimation by Generation and Ordinal Ranking
Saurabh Sharma, Pavan Teja Varigonda, Prashast Bindal, Abhishek Sharma, Arjun Jain
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 CVAE-generated sample set with Oracle selection (vs scoring by predicted ordinals) (ablation) | Ours · Human3.6M Protocol 1 (paired) | MPJPE: 58.0 -> 46.8 | Improved |
| use ground-truth ordinals instead of predicted ordinals for scoring (ablation) | Ours · Human3.6M Protocol 1 (paired) | MPJPE: 58.0 -> 52.1 | Improved |
| use ground-truth ordinals instead of predicted ordinals for scoring (ablation) | Ours · Human3.6M Protocol 2 (paired) | MPJPE: 40.9 -> 36.7 | Improved |
| use ground-truth ordinals instead of predicted ordinals for scoring (ablation) | Ours · HumanEva-I | reconstruction error: 25.9 -> 25.7 | Improved |
| train 2D-to-3D lifting on CMU MoCap without paired images (unpaired) instead of paired Human3.6M | Ours (PRED Ordinals) · Human3.6M Protocol 1 | MPJPE: 58.0 -> 106.8 | Hurt |
| replace Martinez baseline with Ours (PRED Ordinals) in the unpaired setting | Martinez et al. lifting baseline · Human3.6M Protocol 1 (unpaired) | MPJPE: 115.6 -> 106.8 | Improved |
| replace Martinez baseline with Ours (PRED Ordinals) in the unpaired setting | Martinez et al. lifting baseline · Human3.6M Protocol 2 (unpaired) | MPJPE: 71.3 -> 70.5 | Improved |
| increase number of generated samples (ablation) | Oracle selection · Human3.6M Protocol 1 | MPJPE | Improved |
| increase number of generated samples (ablation) | MEAN of samples · Human3.6M Protocol 1 | MPJPE | No clear effect |
| increase number of samples with ordinal scoring (PRED or GT ordinals) (ablation) | OrdinalScore · Human3.6M Protocol 1 | MPJPE | Improved |
| use CVAE MultiPoseNet sampling instead of independent Gaussian Baseline sampling (ablation) | Oracle supervision · Human3.6M Protocol 1 | MPJPE: close to 20mm improvement with more samples | Improved |
| increase Baseline sampling variance to 400mm (ablation) | Baseline sampling · Human3.6M Protocol 1 | MPJPE | Hurt |
Closest papers by meaning
- Ordinal Depth Supervision for 3D Human Pose Estimation (CVPR 2018 · similarity 0.90)
- HMOR: Hierarchical Multi-Person Ordinal Relations for Monocular Multi-Person 3D Pose Estimation (ECCV 2020 · similarity 0.90)
- Probabilistic Monocular 3D Human Pose Estimation With Normalizing Flows (ICCV 2021 · similarity 0.89)
- Monocular 3D Pose Estimation via Pose Grammar and Data Augmentation (TPAMI 2021 · similarity 0.89)
- ManiPose: Manifold-Constrained Multi-Hypothesis 3D Human Pose Estimation (NeurIPS 2024 · similarity 0.88)
- PandaPose: 3D Human Pose Lifting from a Single Image via Propagating 2D Pose Prior to 3D Anchor Space (NeurIPS 2025 · 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.
