Research map / Video / Action recognition
MoAKE: Toward Unified All-in-One Action Quality Assessment via Mixture of Action Knowledge Experts
Huangbiao Xu, Huanqi Wu, Xiao Ke, Jiaxin Cai, Junyi Wu, Jinglin Xu
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 segment-aware temporal aggregation (SATA) (ablation) | single-branch baseline · Ablation (all-in-one, average of 3 datasets) | SRCC: 0.506 -> 0.607 | Improved |
| add AIISRM (ablation) | single-branch baseline with SATA · Ablation (all-in-one, average of 3 datasets) | SRCC: 0.607 -> 0.655 | Improved |
| replace single branch with mixture of action knowledge experts (ablation) | SATA+AIISRM baseline · Ablation (all-in-one, average of 3 datasets) | SRCC: 0.655 -> 0.742 | Improved |
| add shared adapter across experts (ablation) | MoAKE · Ablation (all-in-one, average of 3 datasets) | SRCC: 0.737 -> 0.753 | Improved |
| add AIISRM (ablation) | MoAKE · Ablation (all-in-one, average of 3 datasets) | SRCC: 0.720 -> 0.753 | Improved |
| add SATA and AIISRM (ablation) | MoAKE · Ablation (all-in-one, average of 3 datasets) | SRCC: 0.683 -> 0.753 | Improved |
| add soft router (ablation) | MoAKE · Ablation (all-in-one, average of 3 datasets) | SRCC: 0.529 -> 0.753 | Improved |
| add routing loss L_route (ablation) | MoAKE · Ablation (all-in-one, average of 3 datasets) | SRCC: 0.730 -> 0.753 | Improved |
| add diversity loss L_div (ablation) | MoAKE · Ablation (all-in-one, average of 3 datasets) | SRCC: 0.718 -> 0.753 | Improved |
| add both L_route and L_div (ablation) | MoAKE · Ablation (all-in-one, average of 3 datasets) | SRCC: 0.706 -> 0.753 | Improved |
| train existing methods all-in-one instead of one-by-one | baseline AQA methods · Rhythmic Gymnastics | SRCC: average SRCC drop of 10.17% | Hurt |
| train MoAKE all-in-one instead of one-by-one | MoAKE · RG/Fis-V/LOGO average | SRCC: 0.719 -> 0.753 | Improved |
Closest papers by meaning
- Vision-Language Action Knowledge Learning for Semantic-Aware Action Quality Assessment (ECCV 2024 · similarity 0.90)
- Group-Aware Contrastive Regression for Action Quality Assessment (ICCV 2021 · similarity 0.90)
- Human-Centric Fine-Grained Action Quality Assessment (TPAMI 2025 · similarity 0.89)
- GAIA: Rethinking Action Quality Assessment for AI-Generated Videos (NeurIPS 2024 · similarity 0.89)
- FineParser: A Fine-grained Spatio-temporal Action Parser for Human-centric Action Quality Assessment (CVPR 2024 · similarity 0.88)
- LucidAction: A Hierarchical and Multi-model Dataset for Comprehensive Action Quality Assessment (NeurIPS 2024 · similarity 0.86)
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.
