Research map / 3D vision / Human pose estimation
TexturePose: Supervising Human Mesh Estimation With Texture Consistency
Georgios Pavlakos, Nikos Kolotouros, Kostas Daniilidis
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 texture consistency loss over short monocular clips (ablation) | 2D keypoints + GAN prior · Human3.6M Protocol 1 | reconstruction error (mm): 93.0 -> 80.2 | Improved |
| add texture consistency loss over short monocular clips (ablation) | 2D keypoints + GAN prior · Human3.6M Protocol 2 | reconstruction error (mm): 79.1 -> 76.2 | Improved |
| add unlabeled MPII video frames with texture consistency (ablation) | H36M 3D supervised model · Human3.6M Protocol 1 | reconstruction error (mm): 64.8 -> 60.1 | Improved |
| add unlabeled MPII video frames with texture consistency (ablation) | H36M 3D supervised model · Human3.6M Protocol 2 | reconstruction error (mm): 63.9 -> 58.6 | Improved |
| add MPII 2D keypoint annotations | H36M 3D supervised model · Human3.6M Protocol 1 | reconstruction error (mm): 64.8 -> 54.1 | Improved |
| add unlabeled MPII video frames with texture consistency on top of MPII 2D (ablation) | H36M + MPII 2D · Human3.6M Protocol 1 | reconstruction error (mm): 54.1 -> 51.3 | Improved |
| add unlabeled MPII video frames with texture consistency on top of MPII 2D (ablation) | H36M + MPII 2D · Human3.6M Protocol 2 | reconstruction error (mm): 51.6 -> 49.7 | Improved |
| add texture consistency (vs 3D ground truth for every image) (ablation) | 2D keypoints + GAN prior · Human3.6M Protocol 1 | reconstruction error (mm): 93.0 -> 80.2 | Improved |
| use TexturePose instead of HMR | HMR · Human3.6M Protocol 2 | reconstruction error (mm): 56.8 -> 49.7 | Improved |
| use TexturePose instead of HMR | HMR · LSP part segmentation | accuracy: 87.12 -> 89.00 | Improved |
| use TexturePose instead of HMR | HMR · LSP foreground-background segmentation | accuracy: 91.67 -> 91.82 | Improved |
| use TexturePose instead of Rhodin et al. (second) | Rhodin et al. · Human3.6M Protocol 1 multi-view | MPJPE (mm): 131.7 -> 110.7 | Improved |
Closest papers by meaning
- Convolutional Mesh Regression for Single-Image Human Shape Reconstruction (CVPR 2019 · similarity 0.91)
- Refining 3D Human Texture Estimation From a Single Image (TPAMI 2024 · similarity 0.90)
- Recovering 3D Human Mesh From Monocular Images: A Survey (TPAMI 2023 · similarity 0.89)
- Capturing Humans in Motion: Temporal-Attentive 3D Human Pose and Shape Estimation From Monocular Video (CVPR 2022 · similarity 0.89)
- Neural Localizer Fields for Continuous 3D Human Pose and Shape Estimation (NeurIPS 2024 · similarity 0.88)
- FreeMan: Towards Benchmarking 3D Human Pose Estimation under Real-World Conditions (CVPR 2024 · 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.
