Research map / 3D vision / Human pose estimation
RepNet: Weakly Supervised Training of an Adversarial Reprojection Network for 3D Human Pose Estimation
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 the KCS layer to the critic (ablation) | RepNet · Human3.6M Protocol-II | MPJPE (mm): 77.9 -> 65.1 | Helped |
| Add the KCS layer to the critic (ablation) | RepNet · Human3.6M | symmetry error mean (mm): 31.9 -> 8.2 | Helped |
| Add the KCS layer to the critic (ablation) | RepNet · MPI-INF-3DHP | symmetry error mean (mm): 32.9 -> 11.2 | Helped |
| Use ground-truth 2D labels instead of detected 2D poses | RepNet · Human3.6M Protocol-II | MPJPE (mm): 65.1 -> 38.2 | Helped |
| Train on MPI-INF-3DHP instead of using the Human3.6M-trained model | RepNet · MPI-INF-3DHP | MPJPE (mm): 92.5 -> 97.8 | Hurt |
| Add Gaussian noise with standard deviation 5 to ground-truth 2D joints | RepNet+2DGT · Human3.6M Protocol-II | MPJPE (mm): 38.2 -> 56.9 | Hurt |
| Add Gaussian noise with standard deviation 10 to ground-truth 2D joints | RepNet+2DGT · Human3.6M Protocol-II | MPJPE (mm): 38.2 -> 74.9 | Hurt |
| Add Gaussian noise with standard deviation 15 to ground-truth 2D joints | RepNet+2DGT · Human3.6M Protocol-II | MPJPE (mm): 38.2 -> 89.9 | Hurt |
| Add Gaussian noise with standard deviation 20 to ground-truth 2D joints | RepNet+2DGT · Human3.6M Protocol-II | MPJPE (mm): 38.2 -> 104.6 | Hurt |
Closest papers by meaning
- Neural Reprojection Error: Merging Feature Learning and Camera Pose Estimation (CVPR 2021 · similarity 0.90)
- 3D Human Pose Estimation From a Single Image via Distance Matrix Regression (CVPR 2017 · similarity 0.89)
- 3D Human Pose Estimation in the Wild by Adversarial Learning (CVPR 2018 · similarity 0.89)
- Unsupervised 3D Pose Estimation With Geometric Self-Supervision (CVPR 2019 · similarity 0.89)
- A Dual-Augmentor Framework for Domain Generalization in 3D Human Pose Estimation (CVPR 2024 · similarity 0.87)
- Towards Robust 3D Pose Transfer with Adversarial Learning (CVPR 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.