Research map / Label-efficient and robust learning / Test-time adaptation
Anatomical Domain Shifts: Test-time Heterogeneous Adaptation for 3D Human Pose Prediction
Qiongjie Cui, Pan Zhou, Jingjing Chen, Na Zhao
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 anatomical segment-wise adaptation (TT-HA) instead of full-body HoCoTTA | siMLPe + TTA · H3.6M->GRAB setup-4 (RICH, setup-1) | MPJPE 1000ms: 127.2 -> 119.4 | Improved |
| use TT-HA instead of HoCoTTA | siMLPe + TTA · GRAB unseen categories | MPJPE 1000ms: 136.2 -> 132.2 | Improved |
| use TT-HA instead of HoCoTTA | siMLPe + TTA · GRAB unseen subjects | MPJPE 1000ms: 130.1 -> 127.7 | Improved |
| use TT-HA instead of HoCoTTA | siMLPe + TTA · CMU Mocap setup-1 | MPJPE 1000ms: 71.1 -> 68.5 | Improved |
| use TT-HA instead of HoCoTTA | siMLPe + TTA · GRAB setup-1 | MPJPE 400ms: 45.2 -> 44.3 | No clear effect |
| use TT-HA instead of HoCoTTA | siMLPe + TTA · H3.6M unseen categories | MPJPE 160ms: 22.1 -> 22.4 | Hurt |
| add test-time adaptation (HoCoTTA) to siMLPe | siMLPe · GRAB unseen subjects | MPJPE 1000ms: 155.9 -> 130.1 | Improved |
| set momentum eta to 0.95 instead of 0.92 (ablation) | TT-HA · H3.6M->GRAB setup-4 | MPJPE 1000ms: 121.7 -> 117.1 | Improved |
| set momentum eta to 0.95 instead of 0.98 (ablation) | TT-HA · H3.6M->GRAB setup-4 | MPJPE 1000ms: 120.6 -> 117.1 | Improved |
| set peak threshold tau to 12 instead of 10 (ablation) | TT-HA · H3.6M->GRAB setup-4 | MPJPE 1000ms: 122.0 -> 117.1 | Improved |
| set peak threshold tau to 12 instead of 14 (ablation) | TT-HA · H3.6M->GRAB setup-4 | MPJPE 1000ms: 121.6 -> 117.1 | Improved |
| set window size w to 36 instead of 48 (ablation) | TT-HA · H3.6M->GRAB setup-4 | MPJPE 1000ms: 120.1 -> 117.1 | Improved |
Closest papers by meaning
- Prior-guided Source-free Domain Adaptation for Human Pose Estimation (ICCV 2023 · similarity 0.88)
- Fast Adaptation for Human Pose Estimation via Meta-Optimization (CVPR 2024 · similarity 0.87)
- FlexPose: Pose Distribution Adaptation with Limited Guidance (AAAI 2025 · similarity 0.86)
- Semantics-aware Test-time Adaptation for 3D Human Pose Estimation (ICML 2025 · similarity 0.86)
- PoseBH: Prototypical Multi-Dataset Training Beyond Human Pose Estimation (CVPR 2025 · similarity 0.85)
- A Dual-Augmentor Framework for Domain Generalization in 3D Human Pose Estimation (CVPR 2024 · similarity 0.85)
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.
