Research map / Video / Action recognition
FineDiving: A Fine-Grained Dataset for Procedure-Aware Action Quality Assessment
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 |
|---|---|---|---|
| Replace video-level regression with procedure-aware segmentation, cross-attention, and fine-grained contrastive regression (ablation) | I3D-based action quality assessment · FineDiving (w/ DN) | Spearman's rank correlation: 0.8793 -> 0.9203 | Helped |
| Add procedure segmentation (ablation) | I3D + score regression · FineDiving (w/ DN) | Spearman's rank correlation: 85.76% to 87.93% | Helped |
| Add procedure-aware cross-attention and fine-grained contrastive regression after segmentation (ablation) | I3D + procedure segmentation · FineDiving (w/ DN) | Spearman's rank correlation: 87.93% to 92.03% | Helped |
| Use asymmetric rather than symmetric training (ablation) | I3D + score regression · FineDiving (w/o DN) | Spearman's rank correlation: 0.8504 -> 0.8452 | Hurt |
| Concatenate dive numbers to I3D features (ablation) | I3D + score regression · FineDiving (w/o DN) | Spearman's rank correlation: 0.8504 -> 0.8516 | Helped |
| Use ground-truth step transitions rather than predicted transitions (ablation) | TSA · FineDiving (w/ DN) | Spearman's rank correlation: 0.9203 -> 0.9310 | Helped |
| Increase multi-exemplar voting from 10 exemplars to 15 (ablation) | TSA (w/ DN) · FineDiving | Spearman's rank correlation: 0.9203 -> 0.9204 | No clear effect |
| Use 3 rather than 5 frames in each step (ablation) | TSA (w/ DN) · FineDiving | Spearman's rank correlation: 0.9081 vs 0.9203 | Hurt |
| Use 8 rather than 5 frames in each step (ablation) | TSA (w/ DN) · FineDiving | Spearman's rank correlation: 0.9203 -> 0.9198 | No clear effect |
| Use 5 rather than 3 transformer decoder layers (ablation) | TSA (w/ DN) · FineDiving | Spearman's rank correlation: 0.9203 -> 0.9202 | No clear effect |
Closest papers by meaning
- FineParser: A Fine-grained Spatio-temporal Action Parser for Human-centric Action Quality Assessment (CVPR 2024 · similarity 0.93)
- LucidAction: A Hierarchical and Multi-model Dataset for Comprehensive Action Quality Assessment (NeurIPS 2024 · similarity 0.88)
- FineSports: A Multi-person Hierarchical Sports Video Dataset for Fine-grained Action Understanding (CVPR 2024 · similarity 0.88)
- ActionAtlas: A VideoQA Benchmark for Domain-specialized Action Recognition (NeurIPS 2024 · similarity 0.86)
- ViSTec: Video Modeling for Sports Technique Recognition and Tactical Analysis (AAAI 2024 · similarity 0.86)
- GAIA: Rethinking Action Quality Assessment for AI-Generated Videos (NeurIPS 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.