Research map / Label-efficient and robust learning / Test-time adaptation
Meta-Auxiliary Learning for Adaptive Human Pose Prediction
Read the paperCheck 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 Gated Sharing Units (GSUs) (ablation) | Meta-Auxiliary Learning · H3.6M unseen categories | MPJPE: 11.2 -> 10.1; 23.1 -> 21.6; 60.3 -> 55.2; 109.3 -> 104.1 | Helped |
| Use both Aux.1 and Aux.2 instead of either auxiliary task alone (ablation) | Meta-Auxiliary Learning · H3.6M unseen categories | MPJPE: 12.8 -> 10.1; 11.4 -> 10.1; 25.2 -> 21.6; 22.6 -> 21.6; 62.7 -> 55.2; 58.5 -> 55.2; 111.0 -> 104.1; 108.4 -> 104.1 | Helped |
| Increase test-time gradient updates from 0 to 6 (ablation) | Meta-Auxiliary Learning · H3.6M unseen categories | MPJPE: 10.7 -> 10.1; 23.2 -> 21.6; 58.1 -> 55.2; 109.5 -> 104.1 | Helped |
| Increase test-time gradient updates from 6 to 7 (ablation) | Meta-Auxiliary Learning · H3.6M unseen categories | MPJPE: 10.1 -> 10.7; 21.6 -> 22.0; 55.2 -> 54.9; 104.1 -> 106.7 | Mixed |
| Test-time adapt the forecasting model to the sequence | Meta-Auxiliary Learning · H3.6M S_5, walking/eating/smoking | MPJPE: 56.4 -> 52.3; 76.0 -> 71.6; 69.5 -> 67.5 | Helped |
| Test-time adapt the forecasting model to each unseen subject | Meta-Auxiliary Learning · H3.6M unseen subjects, 1000ms | MPJPE: 113.2 -> 107.0; 127.3 -> 123.2; 124.4 -> 118.7; 116.5 -> 113.5; 114.6 -> 109.7; 112.0 -> 110.2 | Helped |
| Test-time adapt to an unseen action category | Meta-Auxiliary Learning · CMU MoCap unseen categories, 1000ms | MPJPE: 94.9 -> 87.4; 61.9 -> 53.1; 113.0 -> 97.3; 134.8 -> 120.4; 56.3 -> 46.2; 105.1 -> 93.2; 41.4 -> 35.8; 74.7 -> 62.5 | Helped |
Closest papers by meaning
- Fast Adaptation for Human Pose Estimation via Meta-Optimization (CVPR 2024 · similarity 0.91)
- ALIEN: Implicit Neural Representations for Human Motion Prediction under Arbitrary Latency (CVPR 2025 · similarity 0.86)
- LAL: Enhancing 3D Human Motion Prediction with Latency-aware Auxiliary Learning (CVPR 2025 · similarity 0.85)
- MoML: Online Meta Adaptation for 3D Human Motion Prediction (CVPR 2024 · similarity 0.83)
- SimpliHuMoN: Simplifying Human Motion Prediction (NeurIPS 2026 · similarity 0.83)
- MotionMap: Representing Multimodality in Human Pose Forecasting (CVPR 2025 · similarity 0.83)
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.