Research map / Video / Action recognition
Action Quality Assessment with Temporal Parsing Transformer
Yang Bai, Desen Zhou, Songyang Zhang, Jian Wang, Errui Ding, Yu Guan, Yang Long, Jingdong Wang
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 temporal parsing transformer to baseline (ablation) | contrastive regression · MTL-AQA (w/ DD) | Spearman correlation: 0.9498 -> 0.9522 | Improved |
| add ranking loss on attention centers (ablation) | TPT · MTL-AQA (w/ DD) | Spearman correlation: 0.9522 -> 0.9583 | Improved |
| add sparsity loss (ablation) | TPT + ranking loss · MTL-AQA (w/ DD) | Spearman correlation: 0.9583 -> 0.9607 | Improved |
| add sparsity loss (ablation) | TPT + ranking loss · MTL-AQA (w/ DD) | relative L2 (x100): 0.2444 -> 0.2378 | Improved |
| use TPT instead of adaptive pooling for parts (ablation) | part generation · MTL-AQA (w/ DD) | Spearman correlation: 0.9509 -> 0.9607 | Improved |
| use TPT instead of strided temporal convolution for parts (ablation) | part generation · MTL-AQA (w/ DD) | Spearman correlation: 0.9526 -> 0.9607 | Improved |
| use ranking loss instead of diversity loss (ablation) | TPT · MTL-AQA (w/ DD) | Spearman correlation: 0.9538 -> 0.9607 | Improved |
| add position encoding to clip memory and queries (ablation) | TPT · MTL-AQA (w/ DD) | Spearman correlation: 0.9607 -> 0.9526 | Hurt |
| add position encoding to clip memory only (ablation) | TPT · MTL-AQA (w/ DD) | Spearman correlation: 0.9607 -> 0.9532 | Hurt |
| use part-wise relative representation plus AvgPool instead of part-enhanced holistic (ablation) | TPT · MTL-AQA (w/ DD) | Spearman correlation: 0.9578 -> 0.9607 | Improved |
| part-enhanced holistic representation over holistic baseline (ablation) | contrastive regression · MTL-AQA (w/ DD) | Spearman correlation: 0.9498 -> 0.9578 | Improved |
| add transformer encoder to TPT decoder (ablation) | TPT · MTL-AQA | Spearman correlation | Hurt |
Closest papers by meaning
- UIL-AQA: Uncertainty-Aware Clip-Level Interpretable Action Quality Assessment (IJCV 2025 · similarity 0.92)
- Human-Centric Fine-Grained Action Quality Assessment (TPAMI 2025 · similarity 0.90)
- Vision-Language Action Knowledge Learning for Semantic-Aware Action Quality Assessment (ECCV 2024 · similarity 0.90)
- FineParser: A Fine-grained Spatio-temporal Action Parser for Human-centric Action Quality Assessment (CVPR 2024 · similarity 0.90)
- LucidAction: A Hierarchical and Multi-model Dataset for Comprehensive Action Quality Assessment (NeurIPS 2024 · similarity 0.87)
- GAIA: Rethinking Action Quality Assessment for AI-Generated Videos (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.
