Research map / Video / Action recognition
One-Shot Action Recognition via Multi-Scale Spatial-Temporal Skeleton Matching
Siyuan Yang, Jun Liu, Shijian Lu, M.H. Er, Alex Chichung Kot
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 |
|---|---|---|---|
| use optimal-transport (EMD) matching on local features instead of global pooled cosine (ablation) | S-scale · NTU RGB+D 120 (Protocol 2) | accuracy: 60.4 -> 63.2 | Improved |
| use local joint-level features instead of global pooled features (Cosine) (ablation) | S-scale · NTU RGB+D 120 (Protocol 2) | accuracy: 60.4 -> 61.3 | Improved |
| add multi-temporal-scale matching (ablation) | S-scale · NTU RGB+D 120 (Protocol 2) | accuracy: 63.2 -> 65.0 | Improved |
| add multi-spatial-scale matching (ablation) | S-scale · NTU RGB+D 120 (Protocol 2) | accuracy: 63.2 -> 65.1 | Improved |
| add both multi-spatial and multi-temporal matching (ablation) | S-scale · NTU RGB+D 120 (Protocol 2) | accuracy: 63.2 -> 67.6 | Improved |
| add cross-scale matching on top of multi-scale (ablation) | M-scale · NTU RGB+D 120 (Protocol 2) | accuracy: 67.6 -> 68.7 | Improved |
| add cross-scale matching on top of multi-scale (ablation) | M-scale · NTU RGB+D (Protocol 2) | accuracy: 81.6 -> 82.7 | Improved |
| add cross-scale matching on top of multi-scale (ablation) | M-scale · PKU-MMD (Protocol 2) | accuracy: 85.7 -> 86.9 | Improved |
| use scales 1,2,3 instead of scale 1 for multi-spatial matching (ablation) | M-scale · NTU RGB+D 120 (Protocol 2) | accuracy: 63.2 -> 65.1 | Improved |
| use 4 temporal scales instead of 3 (ablation) | multi-temporal matching · NTU RGB+D 120 (Protocol 2) | accuracy: 65.0 -> 64.8 | No clear effect |
| use pooling to min dimension instead of pooling all in cross-scale matching (ablation) | M&C-scale · NTU RGB+D 120 (Protocol 2) | accuracy: 68.7 -> 68.8 | No clear effect |
| use Pooling2_4 instead of Pooling5_10 temporal pooling (ablation) | multi-temporal matching · NTU RGB+D 120 (Protocol 2) | accuracy: 64.0 -> 65.0 | Improved |
Closest papers by meaning
- Spatio-Temporal Difference Descriptor for Skeleton-Based Action Recognition (AAAI 2021 · similarity 0.90)
- Learning Spatial-Preserved Skeleton Representations for Few-Shot Action Recognition (ECCV 2022 · similarity 0.90)
- Revealing Key Details to See Differences: A Novel Prototypical Perspective for Skeleton-based Action Recognition (CVPR 2025 · similarity 0.90)
- Skeleton-Based Action Recognition with Spatial Reasoning and Temporal Stack Learning (ECCV 2018 · similarity 0.90)
- Stitch, Contrast, and Segment: Learning a Human Action Segmentation Model Using Trimmed Skeleton Videos (AAAI 2025 · similarity 0.89)
- Heterogeneous Skeleton-Based Action Representation Learning (CVPR 2025 · similarity 0.89)
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.
