Research map / Label-efficient and robust learning / Recognition and long-tail
D3TW: Discriminative Differentiable Dynamic Time Warping for Weakly Supervised Action Alignment and Segmentation
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 discriminative modeling between positive and negative transcripts (ablation) | D3TW · Breakfast action segmentation | frame accuracy: 38.0 -> 45.7 | Helped |
| Add discriminative modeling between positive and negative transcripts (ablation) | D3TW · Breakfast action segmentation | unit accuracy: 38.4 -> 47.4 | Helped |
| Add discriminative modeling between positive and negative transcripts (ablation) | D3TW · Hollywood Extended action segmentation | frame accuracy: 30.0 -> 33.6 | Helped |
| Add discriminative modeling between positive and negative transcripts (ablation) | D3TW · Hollywood Extended action segmentation | unit accuracy: 28.3 -> 30.5 | Helped |
| Add differentiable relaxation and discriminative modeling instead of the iterative strategy (ablation) | D3TW · Breakfast action segmentation | frame accuracy: 34.9 -> 45.7 | Helped |
| Add differentiable relaxation and discriminative modeling instead of the iterative strategy (ablation) | D3TW · Breakfast action alignment | frame accuracy: 42.8 -> 57.0 | Helped |
| Add differentiable relaxation and discriminative modeling instead of the iterative strategy (ablation) | D3TW · Breakfast action alignment | IoD: 49.5 -> 56.3 | Helped |
| Add discriminative modeling (ablation) | D3TW · Breakfast action alignment | IoD: 47.6 -> 56.3 | Helped |
| Add discriminative modeling (ablation) | D3TW · Hollywood Extended action alignment | frame accuracy: 51.8 -> 59.4 | Helped |
| Add discriminative modeling (ablation) | D3TW · Hollywood Extended action alignment | IoD: 46.9 -> 50.9 | Helped |
| Add discriminative modeling (ablation) | D3TW · Hollywood Extended action segmentation | frame accuracy: 30.0 -> 33.6 | Helped |
Closest papers by meaning
- Representation Learning via Global Temporal Alignment and Cycle-Consistency (CVPR 2021 · similarity 0.91)
- Efficient and Effective Weakly-Supervised Action Segmentation via Action-Transition-Aware Boundary Alignment (CVPR 2024 · similarity 0.89)
- Learning To Align Sequential Actions in the Wild (CVPR 2022 · similarity 0.89)
- Learning by Aligning Videos in Time (CVPR 2021 · similarity 0.88)
- Action Sets: Weakly Supervised Action Segmentation Without Ordering Constraints (CVPR 2018 · similarity 0.88)
- Temporal Alignment-Free Video Matching for Few-shot Action Recognition (CVPR 2025 · similarity 0.87)
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.