Research map / 3D vision / Human pose estimation
AdaptPose: Cross-Dataset Adaptation for 3D Human Pose Estimation by Learnable Motion Generation
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 |
|---|---|---|---|
| adapt generated pose and camera distributions to the target domain | VideoPose3D with AdaptPose · 3DHP test | MPJPE: 96.4 -> 77.2 | Helped |
| add bone generation (ablation) | AdaptPose · 3DHP test | MPJPE: 88.1 -> 77.2 | Helped |
| add camera generation (ablation) | AdaptPose · 3DHP test | MPJPE: 96.4 -> 88.2 | Helped |
| add the domain discriminator for target-domain adaptation (ablation) | AdaptPose with bone and camera generation · 3DHP test | MPJPE: 88.1 -> 78.6 | Helped |
| add sample selection (ablation) | AdaptPose with bone generation, camera generation, and domain discriminator · 3DHP test | MPJPE: 78.6 -> 77.2 | Helped |
| use BG3 axis-angle bone rotation instead of BG1 bone displacement generation (ablation) | AdaptPose · 3DHP test | MPJPE: 85.1 -> 77.2 | Helped |
| use BG3 axis-angle bone rotation instead of BG2 bone generation (ablation) | AdaptPose · 3DHP test | MPJPE: 80.0 -> 77.2 | Helped |
| learn a probabilistic camera distribution instead of one deterministic camera (ablation) | AdaptPose · 3DHP test | MPJPE: 82.8 -> 77.2 | Helped |
| use probabilistic axis-angle rather than quaternion camera representation (ablation) | AdaptPose · 3DHP test | MPJPE: 83.5 -> 77.2 | Helped |
| use probabilistic axis-angle rather than Euler-angle camera representation (ablation) | AdaptPose · 3DHP test | MPJPE: 85.3 -> 77.2 | Helped |
| remove temporal input and use single-frame pairs (ablation) | AdaptPose · 3DHP test | MPJPE: 77.2 -> 86.4 | Hurt |
| use AdaptPose rather than BOA | cross-dataset pose estimation · 3DPW test | MPJPE: 77.2 -> 81.2 | Hurt |
Closest papers by meaning
- FlexPose: Pose Distribution Adaptation with Limited Guidance (AAAI 2025 · similarity 0.90)
- Sim2real transfer learning for 3D human pose estimation: motion to the rescue (NeurIPS 2019 · similarity 0.90)
- CEE-Net: Complementary End-to-End Network for 3D Human Pose Generation and Estimation (AAAI 2023 · similarity 0.90)
- A Dual-Augmentor Framework for Domain Generalization in 3D Human Pose Estimation (CVPR 2024 · similarity 0.89)
- Can Generative Video Models Help Pose Estimation? (CVPR 2025 · similarity 0.88)
- ManiPose: Manifold-Constrained Multi-Hypothesis 3D Human Pose Estimation (NeurIPS 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.