Research map / 3D vision / Human pose estimation
Learning Feature Pyramids for Human Pose Estimation
Wei Yang, Shuang Li, Wanli Ouyang, Hongsheng Li, Xiaogang Wang
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 variance-controlling 1x1 conv (BN-ReLU) in place of identity mapping when summing residual units (ablation) | 2-stack Hourglass · MPII validation | PCKh: 87.6 -> 88.0 | Improved |
| add Pyramid Residual Module on top of variance control (ablation) | 2-stack Hourglass · MPII validation | PCKh: 88.0 -> 88.5 | Improved |
| use PRM-B* pyramid instead of residual unit in hourglass | Stacked Hourglass · MPII test | PCKh@0.5 mean (%): 90.9 -> 92.0 | Improved |
| use PRM Hourglass instead of previous best on LSP | Wei et al. / Bulat&Tzimiropoulos · LSP test | PCK@0.2 mean (%): 90.7 -> 93.9 | Improved |
| use all MPII training set instead of the reduced training set | PRM Hourglass · MPII test | PCKh@0.5 mean (%): 91.8 -> 92.0 | Improved |
| use any PRM design (A to D) instead of plain residual unit (ablation) | 2-stack Hourglass · MPII validation | PCKh | Improved |
| use PRM-A/C/D instead of shared-input PRM-B (ablation) | PRM variants · MPII validation | PCKh | No clear effect |
| widen baseline channels to match model size (Wide BS) versus adding PRM (ablation) | 2-stack Hourglass · MPII validation | PCKh | Mixed |
| increase number of pyramid scales from 3 to 5 at fixed model size (ablation) | PRM-B* · MPII validation | PCKh | Improved |
| use proposed multi-branch initialization instead of Xavier or MSR (ablation) | PRM Hourglass · MPII validation | PCKh | Improved |
| add pyramid branches to Wide ResNet (Ours-28-9 vs WRN-28-10) | WRN-28-10 · CIFAR-10 | top-1 test error (%): 4.17 -> 3.82 | Improved |
| add pyramid branches to Wide ResNet at width 10 | WRN-28-10 · CIFAR-10 | top-1 test error (%): 4.17 -> 3.67 | Improved |
Closest papers by meaning
- Adversarial PoseNet: A Structure-Aware Convolutional Network for Human Pose Estimation (ICCV 2017 · similarity 0.90)
- Cascaded Pyramid Network for Multi-Person Pose Estimation (CVPR 2018 · similarity 0.90)
- Human Pose Estimation via Convolutional Part Heatmap Regression (ECCV 2016 · similarity 0.89)
- End-To-End Learning of Deformable Mixture of Parts and Deep Convolutional Neural Networks for Human Pose Estimation (CVPR 2016 · similarity 0.89)
- A Simple yet Effective Baseline for 3D Human Pose Estimation (ICCV 2017 · similarity 0.88)
- Maximum-Margin Structured Learning with Deep Networks for 3D Human Pose Estimation (IJCV 2016 · 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.
