Research map / Video / Action recognition
Curvature-Guided Task Synergy for Skeleton based Temporal Action Segmentation
Guozhang Li, Xinran Duan, Mei Wang, Lizhi Wang, Hua Huang
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 curvature-guided synergy (CGS) loss (ablation) | LaSA · PKU-MMD (X-sub) | F1@50: 63.6 -> 64.7 | Improved |
| add expert-driven decoupling (EDD) (ablation) | LaSA · PKU-MMD (X-sub) | F1@50: 63.6 -> 64.0 | Improved |
| add both CGS and EDD (ablation) | LaSA · PKU-MMD (X-sub) | F1@50: 63.6 -> 65.5 | Improved |
| add CGS and EDD (ablation) | LaSA · LARa | Acc: 75.3 -> 76.6 | Improved |
| curvature guidance instead of Euclidean distance (ablation) | CGS · LARa | F1@50: 57.8 -> 58.7 | Improved |
| curvature guidance instead of cosine distance (ablation) | CGS · LARa | F1@50: 57.0 -> 58.7 | Improved |
| curvature guidance instead of gradient saliency (ablation) | CGS · LARa | Acc: 74.4 -> 76.2 | Improved |
| add both synergy paths instead of C->L only (ablation) | CGS · LARa | F1@50: 58.2 -> 59.0 | Improved |
| add both synergy paths instead of L->C only (ablation) | CGS · LARa | F1@50: 58.1 -> 59.0 | Improved |
| EDD experts instead of independent encoders (ablation) | EDD · LARa | Acc: 75.7 -> 76.2 | Improved |
| EDD experts instead of pyramid decoupling (ablation) | EDD · LARa | Acc: 75.6 -> 76.2 | Improved |
| use inverted curvature directly as boundary prediction | CGS · LARa | F1@50: 59.0 -> 55.1 | Hurt |
Closest papers by meaning
- Action Recognition Using Rate-Invariant Analysis of Skeletal Shape Trajectories (TPAMI 2016 · similarity 0.89)
- Recovering Complete Actions for Cross-dataset Skeleton Action Recognition (NeurIPS 2024 · similarity 0.87)
- Neuron: Learning Context-Aware Evolving Representations for Zero-Shot Skeleton Action Recognition (CVPR 2025 · similarity 0.85)
- Stitch, Contrast, and Segment: Learning a Human Action Segmentation Model Using Trimmed Skeleton Videos (AAAI 2025 · similarity 0.85)
- Revealing Key Details to See Differences: A Novel Prototypical Perspective for Skeleton-based Action Recognition (CVPR 2025 · similarity 0.85)
- BlockGCN: Redefine Topology Awareness for Skeleton-Based Action Recognition (CVPR 2024 · similarity 0.84)
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.
