Research map / Video / Action recognition
Weakly Supervised Action Learning With RNN Based Fine-To-Coarse Modeling
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 |
|---|---|---|---|
| Add subaction modeling (ablation) | GRU weakly supervised action segmentation · Breakfast | Accuracy (Mof): 22.4 -> 33.3 | Helped |
| Add subaction-count reestimation during iterative training (ablation) | GRU with subactions · Breakfast temporal action segmentation | Accuracy (Mof): 28.8 -> 33.3 | Helped |
| Set subaction counts using ground-truth action lengths (ablation) | GRU with subactions · Breakfast temporal action segmentation | Accuracy (Mof): 33.3 -> 51.3 | Helped |
| Use a recurrent GRU instead of a GMM or MLP for fine-grained classification (ablation) | Weakly supervised subaction model · Breakfast | Accuracy (Mof): GMM: 15.3, 23.3, 26.3, 27.0, 26.5; MLP: 22.4, 24.0, 23.7, 23.1, 20.3; GRU: 25.5, 29.1, 28.6, 29.3, 28.8 | Helped |
| Add the coarse HMM model (ablation) | GRU fine-grained action classifier · Breakfast | Accuracy (Mof): 21.3, 20.1, 23.8, 21.8 -> 25.5, 29.1, 28.6, 29.3 | Helped |
| Add subaction-count reestimation (ablation) | GRU weakly supervised segmentation · Hollywood Ext. temporal action segmentation | Jacc. (IoU): 11.2 -> 11.9 | Helped |
| Add subaction-count reestimation (ablation) | GRU weakly supervised action alignment · Breakfast | Jacc. (IoD): 41.5 -> 47.3 | Helped |
| Add subaction-count reestimation (ablation) | GRU weakly supervised action alignment · Hollywood Ext. | Jacc. (IoD): 45.6 -> 46.3 | Helped |
Closest papers by meaning
- Weakly-Supervised Action Segmentation With Iterative Soft Boundary Assignment (CVPR 2018 · similarity 0.91)
- Large-Scale Weakly-Supervised Pre-Training for Video Action Recognition (CVPR 2019 · similarity 0.90)
- Semi-Weakly-Supervised Learning of Complex Actions From Instructional Task Videos (CVPR 2022 · similarity 0.89)
- Action Sets: Weakly Supervised Action Segmentation Without Ordering Constraints (CVPR 2018 · similarity 0.89)
- SCT: Set Constrained Temporal Transformer for Set Supervised Action Segmentation (CVPR 2020 · similarity 0.89)
- Weakly-supervised Temporal Action Localization by Uncertainty Modeling (AAAI 2021 · 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.