Research map / Video / Action recognition
Action Sets: Weakly Supervised Action Segmentation Without Ordering Constraints
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 |
|---|---|---|---|
| Use a Monte-carlo grammar rather than no grammar (ablation) | Action Sets · Breakfast | frame accuracy: 0.099 -> 0.233 | Helped |
| Use a naive grammar rather than no grammar (ablation) | Action Sets · Breakfast | frame accuracy: 0.099 -> 0.134 | Helped |
| Use a text-based grammar over a Monte-carlo grammar (ablation) | Action Sets · Cooking 2 | midpoint hit: 0.098 -> 0.106 | Helped |
| Use a text-based grammar over a Monte-carlo grammar (ablation) | Action Sets · Breakfast and Hollywood Extended | frame accuracy / Jaccard index: 0.233 -> 0.232; 0.093 -> 0.092 | No clear effect |
| Replace naive mean-length approximation with loss-based approximation (ablation) | Action Sets · Breakfast | frame accuracy: 0.201 -> 0.233 | Helped |
| Add a grammar while retaining the length model (ablation) | Action Sets · Breakfast | frame accuracy: 0.099 -> 0.233 | Helped |
| Add a length model while retaining the grammar (ablation) | Action Sets · Breakfast | frame accuracy: 0.154 -> 0.233 | Helped |
| Provide action sets during inference (ablation) | Action Sets · Breakfast and Hollywood Extended | frame accuracy / Jaccard index: 0.233 -> 0.284; 0.093 -> 0.230 | Helped |
| Provide action sets during inference (ablation) | Action Sets · Cooking 2 | midpoint hit: 0.098 -> 0.102 | No clear effect |
| Split long videos into more training videos with fewer actions per video (ablation) | Action Sets · Cooking 2 | midpoint hit: 0.098 -> 0.121 -> 0.174 | Helped |
| Replace Monte-carlo grammar with manually created grammar (ablation) | Action Sets · Breakfast | frame accuracy: 0.233 -> 0.269 | Helped |
| Replace Monte-carlo grammar with ground-truth grammar (ablation) | Action Sets · Breakfast | frame accuracy: 0.233 -> 0.294 | Helped |
Closest papers by meaning
- SCT: Set Constrained Temporal Transformer for Set Supervised Action Segmentation (CVPR 2020 · similarity 0.94)
- Temporal Action Segmentation From Timestamp Supervision (CVPR 2021 · similarity 0.92)
- Weakly Supervised Actor-Action Segmentation via Robust Multi-Task Ranking (CVPR 2017 · similarity 0.91)
- Set-Supervised Action Learning in Procedural Task Videos via Pairwise Order Consistency (CVPR 2022 · similarity 0.91)
- Efficient and Effective Weakly-Supervised Action Segmentation via Action-Transition-Aware Boundary Alignment (CVPR 2024 · similarity 0.89)
- Unsupervised Action Segmentation via Fast Learning of Semantically Consistent Actoms (AAAI 2024 · similarity 0.89)
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.