Research map / Video / Action recognition
LaDy: Lagrangian-Dynamic Informed Network for Skeleton-based Action Segmentation via Spatial-Temporal Modulation
Haoyu Ji, Xueting Liu, Yu Gao, Wenze Huang, Zhihao Yang, Weihong Ren, Zhiyong Wang, Honghai Liu
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 LDS with spatial modulation (ablation) | DeST-based baseline · PKU-MMD v2 (X-sub) | Acc: 73.6 -> 74.8 | Improved |
| add LDS with temporal modulation (ablation) | DeST-based baseline · PKU-MMD v2 (X-sub) | Acc: 73.6 -> 74.9 | Improved |
| add LDS with temporal modulation (ablation) | DeST-based baseline · PKU-MMD v2 (X-sub) | F1@50: 64.3 -> 65.8 | Improved |
| combine spatial and temporal modulation (STM) (ablation) | LDS · PKU-MMD v2 (X-sub) | Acc: 74.8 -> 75.9 | Improved |
| add temporal modulation to LDS+SM (ablation) | LDS+SM · PKU-MMD v2 (X-sub) | F1@50: 65.7 -> 66.3 | Improved |
| add energy consistency loss to LDS+STM (ablation) | LaDy · PKU-MMD v2 (X-sub) | Acc: 75.9 -> 76.2 | Improved |
| add energy consistency loss to LDS+STM (ablation) | LaDy · PKU-MMD v2 (X-sub) | F1@50: 66.3 -> 67.0 | Improved |
| add ECLoss to LDS+SM (ablation) | LDS+SM · PKU-MMD v2 (X-sub) | Acc: 74.8 -> 74.7 | No clear effect |
| add ECLoss to LDS+TM (ablation) | LDS+TM · PKU-MMD v2 (X-sub) | Acc: 74.9 -> 75.4 | Improved |
| add Lagrangian dynamics modules versus best prior method | LaDy · PKU-MMD v2 (X-view) | F1@50: +5.2% | Improved |
| add Lagrangian dynamics modules versus best prior method | LaDy · PKU-MMD v2 (X-sub) | F1@50: +2.7% | Improved |
| add LDS, STM and ECLoss overhead over baseline (ablation) | LaDy · PKU-MMD v2 | F1@50: 3.5G FLOPs and 0.51M parameters for 2.7% absolute gain | Improved |
Closest papers by meaning
- Skeleton Motion Words for Unsupervised Skeleton-Based Temporal Action Segmentation (ICCV 2025 · similarity 0.89)
- Revealing Key Details to See Differences: A Novel Prototypical Perspective for Skeleton-based Action Recognition (CVPR 2025 · similarity 0.89)
- BlockGCN: Redefine Topology Awareness for Skeleton-Based Action Recognition (CVPR 2024 · similarity 0.88)
- Stitch, Contrast, and Segment: Learning a Human Action Segmentation Model Using Trimmed Skeleton Videos (AAAI 2025 · similarity 0.88)
- Neuron: Learning Context-Aware Evolving Representations for Zero-Shot Skeleton Action Recognition (CVPR 2025 · similarity 0.88)
- Recovering Complete Actions for Cross-dataset Skeleton Action Recognition (NeurIPS 2024 · 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.
