Research map / 3D vision / Human pose estimation
Weakly-Supervised Discovery of Geometry-Aware Representation for 3D Human Pose Estimation
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Inject the learned geometry representation into Regression #3 | Regression #3 · Human3.6M Protocol #1 | MPJPE (mm): Regression #3 53.6 -> 46.3 | Helped |
| Inject the learned geometry representation into Regression #3 | Regression #3 · Human3.6M Protocol #2 | MPJPE (mm): Yang et al. 37.7; Ours + Regression#3 41.6 | Mixed |
| Inject the learned geometry representation into Regression #3 | Regression #3 · Human3.6M Protocol #3 | MPJPE (mm): 53.6 -> 50.3 | Helped |
| Inject the learned geometry representation into Regression #3 | Regression #3 · MPI-INF-3DHP (train Human3.6M) | PCK: 68.4 -> 75.9 | Helped |
| Inject the learned geometry representation into Regression #3 | Regression #3 · MPI-INF-3DHP (train Human3.6M) | AUC: 29.4 -> 36.3 | Helped |
| Use raw images instead of skeleton maps for view synthesis (ablation) | BL geometry-representation framework · Human3.6M Protocol #1 | MPJPE (mm): 49.8mm -> 52.6mm | Hurt |
| Use same-view autoencoding instead of cross-view synthesis (ablation) | BL geometry-representation framework · Human3.6M Protocol #1 | MPJPE (mm): 49.8mm -> 49.9mm | No clear effect |
| Add a view-synthesis geometry representation without representation consistency (ablation) | BL · Human3.6M Protocol #1 | MPJPE (mm): 49.8mm -> 48.2mm | Helped |
| Add virtual-camera data augmentation (ablation) | BL + SG · Human3.6M Protocol #1 | MPJPE (mm) | Helped |
| Add representation consistency constraint (ablation) | Augmented geometry-representation model · Human3.6M Protocol #1 | MPJPE (mm): 47.4mm -> 46.3mm | Helped |
Closest papers by meaning
- Kinematic-Structure-Preserved Representation for Unsupervised 3D Human Pose Estimation (AAAI 2020 · similarity 0.92)
- Self-Supervised 3D Human Pose Estimation via Part Guided Novel Image Synthesis (CVPR 2020 · similarity 0.91)
- Weakly-Supervised 3D Human Pose Learning via Multi-View Images in the Wild (CVPR 2020 · similarity 0.91)
- Geometry-Driven Self-Supervised Method for 3D Human Pose Estimation (AAAI 2020 · similarity 0.91)
- Learning Basis Representation to Refine 3D Human Pose Estimations (AAAI 2019 · similarity 0.90)
- Unsupervised Learning of Category-Level 3D Pose from Object-Centric Videos (CVPR 2024 · similarity 0.88)
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.