Research map / 3D vision / Human pose estimation
Deeply Learned Compositional Models for Human Pose Estimation
Wei Tang, Pei Yu, Ying Wu
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 top-down lateral connections (ablation) | 2-level DLCM · MPII val | mean PCKh@0.5 hard joints | Improved |
| add compositional part supervisions (ablation) | 2-level DLCM · MPII val | mean PCKh@0.5 hard joints | Improved |
| use bone-based part representation instead of conventional cube/K-means types (ablation) | 2-level DLCM · MPII val | mean PCKh@0.5 hard joints | Improved |
| go from 2 to 3 semantic levels (ablation) | DLCM · MPII val | mean PCKh@0.5 hard joints | Improved |
| go from 3 to 4 semantic levels (ablation) | DLCM · MPII val | mean PCKh@0.5 hard joints: 38% more parameters and 27% more GFLOPS | No clear effect |
| 3-level DLCM instead of 8-stack hourglass | Newell et al. · FLIC test | PCK@0.2 total: 98.0 -> 99.0 | Improved |
| 3-level DLCM instead of 8-stack hourglass | Newell et al. · MPII test | PCKh@0.5 total: 90.9 -> 92.3 | Improved |
| 3-level DLCM (5 hourglass) instead of 8-stack hourglass | Newell et al. · model complexity | parameters/GFLOPS: 23.7M -> 15.5M parameters; 41.2 -> 33.6 GFLOPS | Improved |
| 3-level DLCM instead of Yang et al. | Yang et al. · LSP test | PCK@0.2 total: 93.9 -> 95.1 | Improved |
Closest papers by meaning
- Human Pose As Compositional Tokens (CVPR 2023 · similarity 0.90)
- Compositional Human Pose Regression (ICCV 2017 · similarity 0.89)
- End-To-End Learning of Deformable Mixture of Parts and Deep Convolutional Neural Networks for Human Pose Estimation (CVPR 2016 · similarity 0.88)
- Deep Autoencoder for Combined Human Pose Estimation and Body Model Upscaling (ECCV 2018 · similarity 0.87)
- Adversarial PoseNet: A Structure-Aware Convolutional Network for Human Pose Estimation (ICCV 2017 · similarity 0.87)
- Thin-Slicing Network: A Deep Structured Model for Pose Estimation in Videos (CVPR 2017 · 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.
