Research map / 3D vision / Human pose estimation
C3P: Cross-Domain Pose Prior Propagation for Weakly Supervised 3D Human Pose Estimation
Cunlin Wu, Yang Xiao, Boshen Zhang, Mingyang Zhang, Zhiguo Cao, Joey Tianyi Zhou
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 2D-to-3D ray projection loss instead of 3D-to-2D projection (ablation) | C3P (P2P) · ITOP | mean mAP: 78.7 -> 84.5 | Improved |
| use 2D-to-3D ray loss (hands) (ablation) | C3P (P2P) · ITOP | hands mAP: 48.6 -> 63.1 | Improved |
| use 2D-to-3D ray loss (feet) (ablation) | C3P (P2P) · ITOP | feet mAP: 76.5 -> 84.2 | Improved |
| add bone length loss (ablation) | C3P (P2P) · CMU Panoptic | mean error (cm): 6.9 -> 6.1 | Improved |
| add bone length loss (ablation) | C3P (P2P) · CMU Panoptic | mAP: 86.6 -> 89.4 | Improved |
| add body symmetry loss (ablation) | C3P (P2P) · CMU Panoptic | mean error (cm): 6.6 -> 6.1 | Improved |
| add body symmetry loss (ablation) | C3P (P2P) · CMU Panoptic | mAP: 87.5 -> 89.4 | Improved |
| add motion consistency loss (ablation) | C3P (P2P) · CMU Panoptic | mean error (cm): 7.4 -> 6.1 | Improved |
| add motion consistency loss (ablation) | C3P (P2P) · CMU Panoptic | mAP: 85.5 -> 89.4 | Improved |
| train with 2D ground-truth pose instead of estimated (ablation) | C3P (P2P) · CMU Panoptic | mAP: 89.4 -> 92.4 | Improved |
| train with 2D ground-truth pose instead of estimated (ablation) | C3P (PT) · CMU Panoptic | mAP: 93.8 -> 96.0 | Improved |
| train with 2D ground-truth pose instead of estimated (ablation) | C3P (P2P) · CMU Panoptic | mean error (cm): 6.1 -> 4.7 | Improved |
Closest papers by meaning
- Self-Supervised 3D Human Pose Estimation via Part Guided Novel Image Synthesis (CVPR 2020 · similarity 0.89)
- Learning 3D Human Pose from Structure and Motion (ECCV 2018 · similarity 0.89)
- Sequential 3D Human Pose and Shape Estimation From Point Clouds (CVPR 2020 · similarity 0.89)
- Kinematic-Structure-Preserved Representation for Unsupervised 3D Human Pose Estimation (AAAI 2020 · similarity 0.88)
- Towards 3D Human Pose Estimation in the Wild: A Weakly-Supervised Approach (ICCV 2017 · similarity 0.88)
- Weakly-Supervised 3D Human Pose Learning via Multi-View Images in the Wild (CVPR 2020 · 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.
