Research map / Video / Action recognition
Set-Constrained Viterbi for Set-Supervised Action 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 |
|---|---|---|---|
| Use SCV rather than Action Set for set-supervised action segmentation | Action segmentation · Breakfast, Cooking2, Hollywood Ext | benchmark-specific segmentation metric: 23.3 -> 30.2; 10.6 -> 14.5; 9.3 -> 17.7 | Helped |
| Use SCV rather than Action Set for alignment given action sets | Action alignment · Breakfast, Cooking2, Hollywood Ext | benchmark-specific alignment metric: 28.4 -> 40.8; 10.6 -> 15.1; 24.2 -> 35.5 | Helped |
| Add dynamic rather than static HMM parameters (ablation) | SCV · Breakfast, Cooking2, Hollywood Ext | benchmark-specific segmentation metric: 28.7 -> 30.2; 13.7 -> 14.5; 16.1 -> 17.7 | Helped |
| Add regularization rather than removing it (ablation) | SCV · Breakfast, Cooking2, Hollywood Ext | benchmark-specific segmentation metric: 26.5 -> 30.2; 12.3 -> 14.5; 12.6 -> 17.7 | Helped |
| Add non-shared-class distance margins to shared-class affinity regularization (ablation) | SCV n-pair regularization · Breakfast, Cooking2, Hollywood Ext | benchmark-specific segmentation metric: 27.8 -> 30.2; 13.1 -> 14.5; 14.8 -> 17.7 | Helped |
| Use hard instead of soft class assignments for feature distances (ablation) | SCV regularization · Breakfast, Cooking2, Hollywood Ext | benchmark-specific segmentation metric: 29.3 -> 30.2; 14.2 -> 14.5; 16.1 -> 17.7 | Helped |
| Remove legal action-sequence grammar during inference (ablation) | SCV · Breakfast test | Mof: 16.6 -> 12.0 | Hurt |
| Replace Monte Carlo Grammar with Transcript Grammar at inference (ablation) | SCV · Breakfast test | Mof: 30.2 -> 34.7 | Helped |
Closest papers by meaning
- Anchor-Constrained Viterbi for Set-Supervised Action Segmentation (CVPR 2021 · similarity 0.95)
- NeuralNetwork-Viterbi: A Framework for Weakly Supervised Video Learning (CVPR 2018 · similarity 0.90)
- SCT: Set Constrained Temporal Transformer for Set Supervised Action Segmentation (CVPR 2020 · similarity 0.90)
- Efficient and Effective Weakly-Supervised Action Segmentation via Action-Transition-Aware Boundary Alignment (CVPR 2024 · similarity 0.87)
- Hierarchical Vector Quantization for Unsupervised Action Segmentation (AAAI 2025 · similarity 0.85)
- FACT: Frame-Action Cross-Attention Temporal Modeling for Efficient Action Segmentation (CVPR 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.