Research map / 3D vision / Human pose estimation
Learning Basis Representation to Refine 3D Human Pose Estimations
Read the paperCheck 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 |
|---|---|---|---|
| apply basis-representation refinement to Simple baseline outputs | Simple Baseline · H36M, 2D noise=20 | PCK3D avg: 71.73 -> 81.03 | Helped |
| apply refinement to Simple baseline outputs | Simple Baseline · H36M, 2D noise=10 | PCK3D avg: 94.25 -> 94.17 | No clear effect |
| apply refinement to Simple baseline outputs | Simple Baseline · H36M, estimated 2D poses | PCK3D avg: 98.03 -> 98.75 | Helped |
| apply refinement to Matrix baseline outputs | Matrix Baseline · H36M, estimated 2D poses | PCK3D avg: 96.81 -> 97.60 | Helped |
| apply refinement to Simple baseline outputs | Simple Baseline · MPI-INF-3DHP | PCK ALL: 69.4 -> 74.5 | Helped |
| apply refinement to Matrix baseline outputs | Matrix Baseline · MPI-INF-3DHP | PCK ALL: 67.2 -> 72.4 | Helped |
| refine corrupted poses with 14 of 17 joints corrupted | basis refinement · H36M test corrupted poses | PCK3D avg: 67.38 -> 84.89 | Helped |
| use learned convex bases instead of PCA bases (ablation) | refinement of Simple baseline · H36M, 2D noise=20 | PCK3D avg: 73.39 -> 81.03 | Helped |
| use learned convex bases instead of sparse coding bases (ablation) | refinement of Simple baseline · H36M, 2D noise=20 | PCK3D avg: 76.18 -> 81.03 | Helped |
| use learned convex bases instead of sparse coding bases (ablation) | refinement of Simple baseline · H36M, 2D noise=10 | PCK3D avg: 93.97 -> 94.17 | No clear effect |
| use a single dictionary for all actions instead of action-wise dictionaries (1000 bases) (ablation) | basis representation · H36M test reconstruction | MPJPE (mm) avg: 28.95 -> 18.79 | Helped |
| increase action-wise bases from 200 to 1000 (ablation) | basis representation · H36M train reconstruction | MPJPE (mm) avg: 7.89 -> 1.96 | Helped |
Closest papers by meaning
- 3D Human Pose Estimation From a Single Image via Distance Matrix Regression (CVPR 2017 · similarity 0.91)
- Kinematic-Structure-Preserved Representation for Unsupervised 3D Human Pose Estimation (AAAI 2020 · similarity 0.90)
- Weakly-Supervised Discovery of Geometry-Aware Representation for 3D Human Pose Estimation (CVPR 2019 · similarity 0.90)
- ManiPose: Manifold-Constrained Multi-Hypothesis 3D Human Pose Estimation (NeurIPS 2024 · similarity 0.89)
- Self-Supervised 3D Human Mesh Recovery from a Single Image with Uncertainty-Aware Learning (AAAI 2024 · similarity 0.88)
- TokenHMR: Advancing Human Mesh Recovery with a Tokenized Pose Representation (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.