Research map / Video / Action recognition
Spatio-Temporal LSTM with Trust Gates for 3D Human Action Recognition
Jun Liu, Amir Shahroudy, Dong Xu, Gang Wang
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 tree traversal of the skeleton instead of a joint chain (ablation) | ST-LSTM · NTU RGB+D (cross subject) | accuracy: 61.7% -> 65.2% | Improved |
| use tree traversal of the skeleton instead of a joint chain (ablation) | ST-LSTM · NTU RGB+D (cross view) | accuracy: 75.5% -> 76.1% | Improved |
| add trust gate (ablation) | ST-LSTM (Tree Traversal) · NTU RGB+D (cross subject) | accuracy: 65.2% -> 69.2% | Improved |
| add trust gate (ablation) | ST-LSTM (Tree Traversal) · NTU RGB+D (cross view) | accuracy: 76.1% -> 77.7% | Improved |
| add trust gate (ablation) | ST-LSTM (Tree Traversal) · NTU RGB+D (side view samples only) | accuracy: 76.5% -> 81.6% | Improved |
| stack two ST-LSTM layers instead of one (ablation) | ST-LSTM (Tree Traversal) + Trust Gate · NTU RGB+D (cross subject) | accuracy: 65.5% -> 69.2% | Improved |
| stack two ST-LSTM layers instead of one (ablation) | ST-LSTM (Tree Traversal) + Trust Gate · NTU RGB+D (cross view) | accuracy: 77.0% -> 77.7% | Improved |
| use tree traversal instead of joint chain (ablation) | ST-LSTM · SBU Interaction | accuracy: 84.7% -> 88.6% | Improved |
| add trust gate (ablation) | ST-LSTM (Tree) · SBU Interaction | accuracy: 88.6% -> 93.3% | Improved |
| use tree traversal instead of joint chain (ablation) | ST-LSTM · UT-Kinect (leave-one-out protocol) | accuracy: 91.0% -> 92.4% | Improved |
| add trust gate (ablation) | ST-LSTM (Tree) · UT-Kinect (leave-one-out protocol) | accuracy: 92.4% -> 97.0% | Improved |
| feed a larger portion of the test video | ST-LSTM (Tree Traversal) + Trust Gate · NTU RGB+D (cross subject) | accuracy: 13.4% -> 69.2% | Improved |
Closest papers by meaning
- Skeleton-Based Action Recognition Using Spatio-Temporal LSTM Network with Trust Gates (TPAMI 2017 · similarity 0.94)
- Global Context-Aware Attention LSTM Networks for 3D Action Recognition (CVPR 2017 · similarity 0.90)
- Regularizing Long Short Term Memory With 3D Human-Skeleton Sequences for Action Recognition (CVPR 2016 · similarity 0.90)
- Ensemble Deep Learning for Skeleton-Based Action Recognition Using Temporal Sliding LSTM Networks (ICCV 2017 · similarity 0.89)
- Online Human Action Detection Using Joint Classification-Regression Recurrent Neural Networks (ECCV 2016 · similarity 0.89)
- Modeling Temporal Dynamics and Spatial Configurations of Actions Using Two-Stream Recurrent Neural Networks (CVPR 2017 · 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.
