Research map / Video / Video-language understanding
The Pros and Cons: Rank-Aware Temporal Attention for Skill Determination in Long Videos
Hazel Doughty, Walterio Mayol-Cuevas, Dima Damen
Check 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 rank-aware temporal attention instead of Who's Better | Skill ranking · EPIC-Skills | pairwise accuracy: 76.0 -> 80.3 | Improved |
| Use rank-aware temporal attention instead of Who's Better | Skill ranking · BEST | pairwise accuracy: 75.8 -> 81.2 | Improved |
| Add diversity loss to the attention filters (ablation) | Rank-Aware Attention · EPIC-Skills and BEST | pairwise accuracy: +2% | Improved |
| Add rank-aware loss to distinguish high- and low-skill attention (ablation) | Rank-Aware Attention · BEST | pairwise accuracy: 5% | Improved |
| Add rank-aware loss to distinguish high- and low-skill attention (ablation) | Rank-Aware Attention · Scramble Eggs | pairwise accuracy: +10.4% | Improved |
| Add rank-aware loss to distinguish high- and low-skill attention (ablation) | Rank-Aware Attention · Apply Eyeliner | pairwise accuracy: +8.8% | Improved |
| Add disparity loss to make attention outperform uniform weighting (ablation) | Rank-Aware Attention · Drawing | pairwise accuracy | Hurt |
| Add disparity loss to make attention outperform uniform weighting (ablation) | Rank-Aware Attention · Origami | pairwise accuracy | Hurt |
| Use two rank-aware attention modules instead of one rank-agnostic module with six filters (ablation) | Skill ranking on BEST · BEST | pairwise accuracy: 81.2% -> 75.0% | Improved |
| Fuse the high- and low-skill branch scores (ablation) | Rank-Aware Attention · EPIC-Skills | pairwise accuracy: +2.9% | Improved |
| Fuse the high- and low-skill branch scores (ablation) | Rank-Aware Attention · BEST | pairwise accuracy: +3.2% | Improved |
| Use three attention filters instead of one (ablation) | Rank-Aware Attention · Most tasks | pairwise accuracy | Improved |
Closest papers by meaning
- Who's Better? Who's Best? Pairwise Deep Ranking for Skill Determination (CVPR 2018 · similarity 0.90)
- SPAR: A Sequential Primacy and Attribution Ranking Framework for Skill Determination (ECCV 2026 · similarity 0.89)
- SLVMBench: Skill Learning from Video Memory (NeurIPS 2026 · similarity 0.89)
- Masked Temporal Interpolation Diffusion for Procedure Planning in Instructional Videos (ICLR 2025 · similarity 0.85)
- MMBench-Video: A Long-Form Multi-Shot Benchmark for Holistic Video Understanding (NeurIPS 2024 · similarity 0.85)
- Video-STaR: Self-Training Enables Video Instruction Tuning with Any Supervision (ICLR 2025 · similarity 0.84)
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.
