Research map / Video / Action recognition
Weakly Supervised Energy-Based Learning for Action Segmentation
Jun Li, Peng Lei, Sinisa Todorovic
Read the paperCheck 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 |
|---|---|---|---|
| add invalid-path term (L_DF instead of L_F) (ablation) | CDFL, window 20 · Breakfast action segmentation | Mof: 44.3 -> 47.0 | Improved |
| constrain invalid paths to hard ones (L_CDF instead of L_DF) (ablation) | CDFL, window 20 · Breakfast action segmentation | Mof: 47.0 -> 50.2 | Improved |
| use neighbor window of 20 frames instead of single path (window 0) (ablation) | CDFL (L_CDF) · Breakfast action segmentation | Mof: 48.5 -> 50.2 | Improved |
| enlarge window from 20 to 30 (ablation) | CDFL (L_CDF) · Breakfast action segmentation | Mof: 50.2 -> 49.4 | Hurt |
| use alpha 0.1 instead of 0 in L_DF (ablation) | L_DF, window 20 · Breakfast action segmentation | Mof: 44.3 -> 47.0 | Improved |
| raise alpha from 0.1 to 0.2 in L_DF (ablation) | L_DF, window 20 · Breakfast action segmentation | Mof: 47.0 -> 40.7 | Hurt |
| raise alpha from 0.1 to 0.3 in L_DF (ablation) | L_DF, window 20 · Breakfast action segmentation | Mof: 47.0 -> 35.5 | Hurt |
| add invalid-path term (L_DF instead of L_F) (ablation) | CDFL, window 6 · Hollywood Extended action alignment | IoD: 49.3 -> 50.5 | Improved |
| constrain invalid paths to hard ones (L_CDF instead of L_DF) (ablation) | CDFL, window 6 · Hollywood Extended action alignment | IoD: 50.5 -> 52.9 | Improved |
| use neighbor window of 6 instead of window 0 (ablation) | CDFL (L_CDF) · Hollywood Extended action alignment | IoD: 49.8 -> 52.9 | Improved |
| enlarge window from 6 to 8 (ablation) | CDFL (L_CDF) · Hollywood Extended action alignment | IoD: 52.9 -> 51.6 | Hurt |
Closest papers by meaning
- A Hybrid RNN-HMM Approach for Weakly Supervised Temporal Action Segmentation (TPAMI 2018 · similarity 0.91)
- Action Sets: Weakly Supervised Action Segmentation Without Ordering Constraints (CVPR 2018 · similarity 0.90)
- Weakly Supervised Action Learning With RNN Based Fine-To-Coarse Modeling (CVPR 2017 · 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 0.88)
- FACT: Frame-Action Cross-Attention Temporal Modeling for Efficient Action Segmentation (CVPR 2024 · similarity 0.88)
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.
