Research map / 3D vision / Human pose estimation
Single-Network Whole-Body Pose Estimation
Gines Hidalgo, Yaadhav Raaj, Haroon Idrees, Donglai Xiang, Hanbyul Joo, Tomas Simon, Yaser Sheikh
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 |
|---|---|---|---|
| use deeper higher-resolution architecture instead of shallow OpenPose-sized network (ablation) | whole-body network · COCO val (multi-scale) | Body AP: 60.9 -> 65.6 | Improved |
| use deeper higher-resolution architecture instead of shallow network (ablation) | whole-body network · COCO val (multi-scale) | Foot AP: 70.2 -> 76.2 | Improved |
| add face and hand keypoints to the shallow body-foot model | shallow OpenPose-style network · COCO val body | Body AP: 65.3 -> 60.9 | Hurt |
| add face and hand keypoints to the deep body-foot model | deep network · COCO val body | Body AP: 66.4 -> 65.6 | Hurt |
| deep body-foot architecture instead of OpenPose body-foot | body-foot detector · COCO val (multi-scale) | Body AP: 65.3 -> 66.4 | Improved |
| choose 3 PAF stages with 8 blocks and 96-256w over 1 stage with 10 blocks at equal run-time (ablation) | PAF network · COCO val body | AP: 65.8 -> 66.4 | Improved |
| use 5 PAF stages with narrow width instead of 3 stages (ablation) | PAF network · COCO val body | AP: 66.4 -> 65.5 | Hurt |
| deep instead of shallow whole-body on hard in-the-wild faces (ablation) | whole-body network · i-bug face | Face AR: 50.6 -> 54.5 | Improved |
| deep whole-body vs OpenPose on in-the-wild faces | whole-body network · i-bug face | Face AR: 52.4 -> 54.5 | Improved |
| single-network deep whole-body vs OpenPose on lab-recorded faces | whole-body network · Multi-PIE face | Face AR: 96.3 -> 93.2 | Hurt |
| deep instead of shallow whole-body on hand (ablation) | whole-body network · Hand MPII | Hand AR: 82.4 -> 88.1 | Improved |
| deep whole-body vs OpenPose on in-the-wild hands | whole-body network · Hand MPII | Hand AR: 82.7 -> 88.1 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- OpenPose — OpenPose: Realtime Multi-Person 2D Pose Estimation Using Part Affinity Fields (TPAMI 2019) · better on 4 of 5 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- ZoomNAS: Searching for Whole-body Human Pose Estimation in the Wild (TPAMI 2022 · similarity 0.91)
- Whole-Body Human Pose Estimation in the Wild (ECCV 2020 · similarity 0.91)
- AlphaPose: Whole-Body Regional Multi-Person Pose Estimation and Tracking in Real-Time (TPAMI 2022 · similarity 0.90)
- DOPE: Distillation Of Part Experts for whole-body 3D pose estimation in the wild (ECCV 2020 · similarity 0.90)
- Towards Robust and Expressive Whole-body Human Pose and Shape Estimation (NeurIPS 2023 · similarity 0.90)
- Neural Localizer Fields for Continuous 3D Human Pose and Shape Estimation (NeurIPS 2024 · 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.
