Research map / Video / Action recognition
Ensemble Deep Learning for Skeleton-Based Action Recognition Using Temporal Sliding LSTM Networks
Inwoong Lee, Doyoung Kim, Seoungyoon Kang, Sanghoon Lee
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 human cognitive coordinate transform to LSTM (ablation) | LSTM · MSR Action3D | average accuracy (%): 71.33 -> 77.43 | Improved |
| add salient motion features on top of HCC (ablation) | LSTM + HCC · MSR Action3D | average accuracy (%): 77.43 -> 91.78 | Improved |
| add multi-term TS-LSTM ensemble (v1) on top of HCC and SMF (ablation) | LSTM + HCC + SMF · MSR Action3D | average accuracy (%): 91.78 -> 96.63 | Improved |
| add pose TS-LSTM part and its loss (v2 over v1) (ablation) | Ensemble TS-LSTM · MSR Action3D | average accuracy (%): 96.63 -> 97.22 | Improved |
| add human cognitive coordinate transform to LSTM (ablation) | LSTM · UTKinect-Action | accuracy (%): 60.61 -> 72.73 | Improved |
| add salient motion features on top of HCC (ablation) | LSTM + HCC · UTKinect-Action | accuracy (%): 72.73 -> 93.94 | Improved |
| add multi-term TS-LSTM ensemble (v1) on top of HCC and SMF (ablation) | LSTM + HCC + SMF · UTKinect-Action | accuracy (%): 93.94 -> 95.96 | Improved |
| add pose TS-LSTM part and its loss (v2 over v1) (ablation) | Ensemble TS-LSTM · UTKinect-Action | accuracy (%): 95.96 -> 96.97 | Improved |
| add pose TS-LSTM part and its loss (v2 over v1) (ablation) | Ensemble TS-LSTM · NTU RGB+D (cross-subject) | accuracy (%): 73.88 -> 74.60 | Improved |
| add pose TS-LSTM part and its loss (v2 over v1) (ablation) | Ensemble TS-LSTM · NTU RGB+D (cross-view) | accuracy (%): 80.40 -> 81.25 | Improved |
| add pose TS-LSTM part and its loss (v2 over v1) (ablation) | Ensemble TS-LSTM · Northwestern-UCLA | accuracy (%): 85.99 -> 89.22 | Improved |
| add pose TS-LSTM part and its loss (v2 over v1) (ablation) | Ensemble TS-LSTM · UWA3DII (cross-view mean) | accuracy (%): 72.4 -> 75.6 | Improved |
Closest papers by meaning
- Co-Occurrence Feature Learning for Skeleton Based Action Recognition Using Regularized Deep LSTM Networks (AAAI 2016 · similarity 0.94)
- Skeleton-Based Action Recognition Using Spatio-Temporal LSTM Network with Trust Gates (TPAMI 2017 · similarity 0.92)
- Regularizing Long Short Term Memory With 3D Human-Skeleton Sequences for Action Recognition (CVPR 2016 · similarity 0.92)
- An Attention Enhanced Graph Convolutional LSTM Network for Skeleton-Based Action Recognition (CVPR 2019 · similarity 0.91)
- Bayesian Graph Convolution LSTM for Skeleton Based Action Recognition (ICCV 2019 · similarity 0.90)
- Revealing Key Details to See Differences: A Novel Prototypical Perspective for Skeleton-based Action Recognition (CVPR 2025 · similarity 0.87)
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.
