Research map / Video / Action recognition
Spectral Scalpel: Amplifying Adjacent Action Discrepancy via Frequency-Selective Filtering for Skeleton-Based Action Segmentation
Haoyu Ji, Bowen Chen, Zhihao Yang, Wenze Huang, Yu Gao, Xueting Liu, Weihong Ren, Zhiyong Wang, Honghai Liu
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 MASF to the baseline (ablation) | DeST-based STAS baseline · PKU-MMD v2 (X-sub) | F1@50: 64.3 -> 65.6 | Improved |
| add AADL to the baseline (ablation) | DeST-based STAS baseline · PKU-MMD v2 (X-sub) | F1@50: 64.3 -> 65.5 | Improved |
| add FACM to the baseline (ablation) | DeST-based STAS baseline · PKU-MMD v2 (X-sub) | F1@50: 64.3 -> 65.7 | Improved |
| combine all three modules (MASF+AADL+FACM) (ablation) | DeST-based STAS baseline · PKU-MMD v2 (X-sub) | F1@50: 64.3 -> 66.6 | Improved |
| add FACM on top of MASF+AADL (ablation) | Spectral Scalpel · PKU-MMD v2 (X-sub) | F1@50: 66.0 -> 66.6 | Improved |
| add AADL (training-only loss) on top of MASF+FACM (ablation) | Spectral Scalpel · PKU-MMD v2 (X-sub) | F1@50: 66.2 -> 66.6 | Improved |
| add MASF on top of AADL+FACM (ablation) | Spectral Scalpel · PKU-MMD v2 (X-sub) | F1@50: 66.3 -> 66.6 | Improved |
| Spectral Scalpel vs prior best STAS (LaSA/LPL) | STAS methods · PKU-MMD v2 (X-view) | F1@50: 62.4 -> 67.2 | Improved |
| Spectral Scalpel vs LaSA on MCFS-130 edit score | STAS methods · MCFS-130 | Edit: 79.3 -> 78.0 | Mixed |
| Spectral Scalpel vs prior best on LARa | STAS methods · LARa | F1@50: 58.6 -> 59.4 | Improved |
Closest papers by meaning
- LaDy: Lagrangian-Dynamic Informed Network for Skeleton-based Action Segmentation via Spatial-Temporal Modulation (CVPR 2026 · similarity 0.87)
- Skeleton Motion Words for Unsupervised Skeleton-Based Temporal Action Segmentation (ICCV 2025 · similarity 0.87)
- Stitch, Contrast, and Segment: Learning a Human Action Segmentation Model Using Trimmed Skeleton Videos (AAAI 2025 · similarity 0.86)
- Recovering Complete Actions for Cross-dataset Skeleton Action Recognition (NeurIPS 2024 · similarity 0.84)
- Revealing Key Details to See Differences: A Novel Prototypical Perspective for Skeleton-based Action Recognition (CVPR 2025 · similarity 0.84)
- Neuron: Learning Context-Aware Evolving Representations for Zero-Shot Skeleton Action Recognition (CVPR 2025 · 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.
