Research map / Graph learning / Knowledge and temporal graphs
SkateFormer: Skeletal-Temporal Transformer for Human Action Recognition
Jeonghyeok Do, Munchurl Kim
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 four-type partition-specific Skate-MSA instead of no attention (ablation) | SkateFormer · NTU RGB+D X-Sub60 | top-1 accuracy: 90.7 -> 92.6 | Improved |
| use naive full self-attention instead of partition-specific attention (ablation) | SkateFormer · NTU RGB+D X-View60 | top-1 accuracy: 95.7 -> 95.0 | Hurt |
| Skate-MSA instead of naive self-attention reduces memory (ablation) | SkateFormer · NTU RGB+D | memory (MB): 1612.1 -> 137.6 | Improved |
| add skeletal relation partitions only (ablation) | SkateFormer · NTU RGB+D X-Sub60 | top-1 accuracy: 90.7 -> 91.8 | Improved |
| add temporal relation partitions only (ablation) | SkateFormer · NTU RGB+D X-Sub60 | top-1 accuracy: 90.7 -> 91.9 | Improved |
| add learnable skeletal embedding with fixed temporal embedding (Skate-Embedding) (ablation) | SkateFormer · NTU RGB+D X-Sub60 | top-1 accuracy: 91.9 -> 92.6 | Improved |
| make temporal embedding learnable instead of fixed (ablation) | Skate-Embedding · NTU RGB+D X-Sub60 | top-1 accuracy: 92.6 -> 91.4 | Hurt |
| use trimmed-uniform random sampling instead of fixed-stride (ablation) | SkateFormer · NTU RGB+D X-Sub60 | top-1 accuracy: 91.4 -> 92.6 | Improved |
| add skeletal intra-instance augmentation (ablation) | SkateFormer · NTU RGB+D X-Sub60 | top-1 accuracy: 89.8 -> 92.2 | Improved |
| add inter-instance bone-length exchange augmentation alone (ablation) | SkateFormer · NTU RGB+D X-View60 | top-1 accuracy: 94.6 -> 94.3 | Hurt |
| add inter-instance augmentation on top of intra-instance augmentation (ablation) | SkateFormer · NTU RGB+D X-Sub60 | top-1 accuracy: 92.2 -> 92.6 | Improved |
| increase ensemble from joint-only to four modalities | SkateFormer · NTU RGB+D X-Sub60 | top-1 accuracy: 92.6 -> 93.5 | Improved |
Later papers that compared with this method
1 later accepted paper reports SkateFormer as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- Adaptive Hyper-Graph Convolution Network for Skeleton-based Human Action Recognition with Virtual Connections (ICCV 2025 · better on 3 of 5 shared results)
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- InfoGCN: Representation Learning for Human Skeleton-Based Action Recognition (CVPR 2022 · similarity 0.91)
- Graph CNNs with Motif and Variable Temporal Block for Skeleton-Based Action Recognition (AAAI 2019 · similarity 0.91)
- 3Mformer: Multi-Order Multi-Mode Transformer for Skeletal Action Recognition (CVPR 2023 · similarity 0.91)
- Revealing Key Details to See Differences: A Novel Prototypical Perspective for Skeleton-based Action Recognition (CVPR 2025 · similarity 0.90)
- BlockGCN: Redefine Topology Awareness for Skeleton-Based Action Recognition (CVPR 2024 · similarity 0.90)
- Spatio-Temporal Fusion for Human Action Recognition via Joint Trajectory Graph (AAAI 2024 · 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.
