Research map / 3D vision / Human pose estimation
Integral Human Pose Regression
Xiao Sun, Bin Xiao, Fangyin Wei, Shuang Liang, Yichen Wei
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 |
|---|---|---|---|
| Replace heat map argmax with integral regression (add integral loss on joints) | H1, ResNet-50 · MPII val | PCKh@0.1: 17.2 -> 29.3 | Improved |
| Replace heat map argmax with integral regression | H1, ResNet-50 · MPII val | PCKh@0.5: 86.8 -> 87.3 | Improved |
| Add integral regression with binary cross-entropy heat map loss | H3, ResNet-50 · MPII val | AUC: 46.3 -> 57.7 | Improved |
| Integral regression without heat map supervision (I*) instead of direct regression | R1, ResNet-50 · MPII val | PCKh@0.1: 25.0 -> 28.3 | Improved |
| Add heat map supervision to integral regression (I1 vs I*) (ablation) | Integral regression, ResNet-50 · MPII val | PCKh@0.5: 86.0 -> 87.3 | Improved |
| Use integral regression when halving heat map size at image size 128 (ablation) | H1 vs I1 · MPII val | AUC: H1 drops 12.5, I1 drops 3.5 | Improved |
| Use integral regression with ResNet-18 backbone (ablation) | H1 ResNet-101 · MPII val | PCKh@0.5: I1 ResNet-18 86.0 vs H1 ResNet-101 87.3 | No clear effect |
| Add integral regression to 4-stage network (ablation) | MS-H1 · MPII val | AUC: 54.0 -> 60.1 | Improved |
| Add integral regression to ResNet-101 pose model | H1 · COCO test-dev | AP: 66.3 -> 67.8 | Improved |
| Add integral regression with HourGlass backbone | HG-H1 · MPII test | PCKh@0.5: 90.4 -> 91.0 | Improved |
| Add integral regression with H2 loss, H36M only | H2 · Human3.6M Strategy 1 | MPJPE: 80.4 -> 66.2 | Improved |
| Add integral regression with direct-regression-like I* net, H36M only | R1 · Human3.6M Strategy 1 | MPJPE: 106.6 -> 100.2 | Improved |
Closest papers by meaning
- Removing the Bias of Integral Pose Regression (ICCV 2021 · similarity 0.90)
- Dive Deeper Into Integral Pose Regression (ICLR 2022 · similarity 0.90)
- Bias-Compensated Integral Regression for Human Pose Estimation (TPAMI 2023 · similarity 0.89)
- Continuous Heatmap Regression for Pose Estimation via Implicit Neural Representation (NeurIPS 2024 · similarity 0.86)
- Video-Based Human Pose Regression via Decoupled Space-Time Aggregation (CVPR 2024 · similarity 0.85)
- MotionMap: Representing Multimodality in Human Pose Forecasting (CVPR 2025 · similarity 0.83)
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.
