Research map / Architectures and efficiency / CNN design
Deep Learning on Lie Groups for Skeleton-Based Action Recognition
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 RotMap/RotPooling blocks, increasing from 0 to 3 (ablation) | LieNet · G3D-Gaming | accuracy: 84.55% -> 89.10% | Helped |
| Add a fourth RotMap/RotPooling block instead of using 3 blocks (ablation) | LieNet · G3D-Gaming | accuracy: 89.10% -> 87.28% | Hurt |
| Add RotMap/RotPooling blocks, increasing from 0 to 2 (ablation) | LieNet · HDM05 | accuracy: 71.26%$\pm$2.12 -> 75.78%$\pm$2.26 | Helped |
| Add a third RotMap/RotPooling block instead of using 2 blocks (ablation) | LieNet · HDM05 | accuracy: 75.78%$\pm$2.26 -> 70.42% | Hurt |
| Add RotMap/RotPooling blocks, increasing from 0 to 3 (ablation) | LieNet · NTU RGB+D cross-subject | accuracy: 53.54% -> 61.37% | Helped |
| Add RotMap/RotPooling blocks, increasing from 0 to 3 (ablation) | LieNet · NTU RGB+D cross-view | accuracy: 54.78% -> 66.95% | Helped |
| Add a ReLU-like layer after LogMap (ablation) | LieNet-3Blocks · G3D-Gaming | accuracy: 89.10% -> 87.58% | Hurt |
| Use DTW with RotMap layers (ablation) | LieNet-3Blocks · G3D-Gaming | accuracy: 88.89% vs. 89.10% | No clear effect |
| Use two spatial RotPooling steps rather than only the first spatial pooling step (ablation) | LieNet · G3D-Gaming | accuracy | Hurt |
| Replace LieNet's RotMap/RotPooling layers with a regular LeNet-like stack (ablation) | LieNet-3Blocks · G3D-Gaming | accuracy: 89.10% -> 85.49% | Hurt |
Closest papers by meaning
- Rolling Rotations for Recognizing Human Actions From 3D Skeletal Data (CVPR 2016 · similarity 0.89)
- A New Representation of Skeleton Sequences for 3D Action Recognition (CVPR 2017 · similarity 0.87)
- Novel Motion Patterns Matter for Practical Skeleton-Based Action Recognition (AAAI 2023 · similarity 0.86)
- Modeling Temporal Dynamics and Spatial Configurations of Actions Using Two-Stream Recurrent Neural Networks (CVPR 2017 · similarity 0.86)
- BlockGCN: Redefine Topology Awareness for Skeleton-Based Action Recognition (CVPR 2024 · similarity 0.85)
- Revealing Key Details to See Differences: A Novel Prototypical Perspective for Skeleton-based Action Recognition (CVPR 2025 · similarity 0.85)
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.