Research map / Video / Action recognition
Skeleton-Based Gesture Recognition Using Several Fully Connected Layers with Path Signature Features and Temporal Transformer Module
Read the paperCheck 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 path-signature features (T_PS, S_PS, or T_S_PS) to raw coordinates (ablation) | 1s_net_w/o TTM · ChaLearn 2013 validation | accuracy rate (%): 77.98; 82.30; 85.35 | Helped |
| Combine all three PS feature types with raw coordinates (ablation) | 1s_net_w/o TTM · ChaLearn 2013 validation | accuracy rate (%): 85.50; 88.76; 89.17; 89.20 | Helped |
| Increase PS truncation levels (ablation) | 1s_net_w/o TTM with PS features · ChaLearn 2013 validation | accuracy rate (%): 77.98; 81.28; 84.42; 85.27 | Mixed |
| Add temporal enhancement by shifting samples in [-5, 5] (ablation) | 1s_net_w/o TTM · ChaLearn 2013 validation | accuracy rate (%): 77.84; 77.46 | No clear effect |
| Add TTM to the RC-input one-stream network (ablation) | 1s_net · ChaLearn 2013 validation | accuracy rate (%): 77.84; 80.55 | Helped |
| Add temporal enhancement to the TTM network (ablation) | 1s_net_TTM · ChaLearn 2013 validation | accuracy rate (%): 80.55; 81.33 | Helped |
| Separate RC and PS into two streams (ablation) | 1s_net · ChaLearn 2013 validation | accuracy rate (%): 89.43; 89.63 | Helped |
| Separate spatial and temporal information into three streams (ablation) | 1s_net · ChaLearn 2013 validation | accuracy rate (%): 89.43; 90.19 | Helped |
| Use the proposed three-stream network with TTM instead of the strongest listed baseline | Skeleton-based gesture recognition · ChaLearn 2013 | accuracy rate (%): 91.70; 92.08 | Helped |
| Use the proposed three-stream network with TTM instead of SkeLSTM | Skeleton-based gesture recognition · ChaLearn 2016 test | test accuracy (%): 35.39; 39.95 | Helped |
| Use the proposed three-stream network with TTM instead of MP | Skeleton-based gesture recognition · MSRC-12 | accuracy rate (%): 98.25; 99.01 | Helped |
Closest papers by meaning
- Doodle to Detect: A Goofy but Powerful Approach to Skeleton-based Hand Gesture Recognition (NeurIPS 2025 · similarity 0.90)
- A Neural Network Based on SPD Manifold Learning for Skeleton-Based Hand Gesture Recognition (CVPR 2019 · similarity 0.88)
- Spatio-Temporal Difference Descriptor for Skeleton-Based Action Recognition (AAAI 2021 · similarity 0.88)
- Modeling Temporal Dynamics and Spatial Configurations of Actions Using Two-Stream Recurrent Neural Networks (CVPR 2017 · similarity 0.88)
- Skeleton-based Action Recognition with Non-linear Dependency Modeling and Hilbert-Schmidt Independence Criterion (AAAI 2025 · similarity 0.85)
- Revealing Key Details to See Differences: A Novel Prototypical Perspective for Skeleton-based 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.