Research map / 3D vision / Human pose estimation
Human Pose Estimation via Convolutional Part Heatmap Regression
Adrian Bulat, Georgios Tzimiropoulos
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 regression subnetwork on stacked part heatmaps after the detector (ablation) | Residual part detection · MPII val | PCKh total: 82.6 -> 88.2 | Improved |
| add part-detection heatmap stage before regression instead of direct regression only (ablation) | Residual regression · MPII val | PCKh total: 84.2 -> 88.2 | Improved |
| use detection-followed-by-regression instead of regression-followed-by-regression (residual) (ablation) | Residual cascade · MPII val | PCKh total: 85.7 -> 88.2 | Improved |
| use detection-followed-by-regression instead of regression-followed-by-regression (VGG) (ablation) | VGG-FCN cascade · MPII val | PCKh total: 76.7 -> 83.2 | Improved |
| use residual (ResNet-152 + hourglass) instead of VGG-FCN (ablation) | Part heatmap regression · MPII val | PCKh total: 83.2 -> 88.2 | Improved |
| use residual instead of VGG-FCN | Part heatmap regression · MPII test | PCKh total: 82.7 -> 89.7 | Improved |
| use residual instead of VGG-FCN | Part heatmap regression · LSP | PCK total: 83.5 -> 90.7 | Improved |
| stop backpropagation for missing keypoints instead of generating points artificially when fine-tuning on LSP (ablation) | Part heatmap regression · LSP | PCK | Improved |
| train detector and regressor jointly from the start instead of staged training (ablation) | Part heatmap regression · MPII | training speed | No clear effect |
| different initialization strategies for residual regression network (ablation) | Residual regression · MPII | accuracy | No clear effect |
| compare to Newell et al. on MPII test | Part heatmap regression (Res) · MPII test | PCKh total: 89.4 -> 89.7 | Improved |
Closest papers by meaning
- Adversarial PoseNet: A Structure-Aware Convolutional Network for Human Pose Estimation (ICCV 2017 · similarity 0.90)
- End-To-End Learning of Deformable Mixture of Parts and Deep Convolutional Neural Networks for Human Pose Estimation (CVPR 2016 · similarity 0.89)
- Convolutional Pose Machines (CVPR 2016 · similarity 0.89)
- Learning Feature Pyramids for Human Pose Estimation (ICCV 2017 · similarity 0.89)
- A Simple yet Effective Baseline for 3D Human Pose Estimation (ICCV 2017 · similarity 0.89)
- HEMlets Pose: Learning Part-Centric Heatmap Triplets for Accurate 3D Human Pose Estimation (ICCV 2019 · similarity 0.89)
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.
