Research map / Video / Action recognition
Joint Self-Supervised Video Alignment and Action Segmentation
Ali Shah Ali, Syed Ahmed Mahmood, Mubin Saeed, Andrey Konin, M. Zeeshan Zia, Quoc-Huy Tran
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 structural prior (GW term) (ablation) | VAOT · IKEA ASM | Acc@1.0: 37.81 -> 38.64 | Improved |
| add temporal prior (ablation) | VAOT · IKEA ASM | Acc@1.0: 17.84 -> 38.64 | Improved |
| use balanced instead of full unbalanced assignment (ablation) | VAOT · IKEA ASM | Acc@1.0: 25.24 -> 38.64 | Improved |
| add virtual frame for background frames (ablation) | VAOT · IKEA ASM | Acc@1.0: 36.10 -> 38.64 | Improved |
| add structural prior (GW term) (ablation) | VAOT · Pouring | Kendall tau: 72.49 -> 88.28 | Improved |
| add temporal prior (ablation) | VAOT · Pouring | Kendall tau: 78.11 -> 88.28 | Improved |
| use balanced instead of full unbalanced assignment (ablation) | VAOT · Pouring | Kendall tau: 68.78 -> 88.28 | Improved |
| add virtual frame for background frames (ablation) | VAOT · Pouring | Kendall tau: 82.65 -> 88.28 | Improved |
| use VAOT (FGW with structural prior) instead of VAVA (Kantorovich with optimality prior) | video alignment · IKEA ASM | Acc@1.0: 29.10 -> 38.64 | Improved |
| add action segmentation jointly (VASOT instead of VAOT) | video alignment · IKEA ASM | Acc@1.0: 38.64 -> 31.02 | Hurt |
| add video alignment jointly (VASOT instead of ASOT) | action segmentation · 50 Salads Eval | MoF: 59.3 -> 60.6 | Improved |
| add video alignment jointly (VASOT instead of ASOT) | action segmentation · Desktop Assembly | F1: 68.0 -> 75.1 | Improved |
Closest papers by meaning
- Temporally Consistent Unbalanced Optimal Transport for Unsupervised Action Segmentation (CVPR 2024 · similarity 0.90)
- Efficient and Effective Weakly-Supervised Action Segmentation via Action-Transition-Aware Boundary Alignment (CVPR 2024 · similarity 0.87)
- OPEL: Optimal Transport Guided ProcedurE Learning (NeurIPS 2024 · similarity 0.86)
- Learning to Predict Activity Progress by Self-Supervised Video Alignment (CVPR 2024 · similarity 0.85)
- Temporal Alignment-Free Video Matching for Few-shot Action Recognition (CVPR 2025 · similarity 0.85)
- Align Before Adapt: Leveraging Entity-to-Region Alignments for Generalizable Video Action Recognition (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.
