Research map / Video / Action recognition
Regularizing Long Short Term Memory With 3D Human-Skeleton Sequences for 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 class-independent 3D-skeleton feature constraints (ablation) | GoogLeNet + LSTM · UCF101 | average classification accuracy: 75.2 -> 78.3 | Helped |
| add class-independent 3D-skeleton feature constraints (ablation) | GoogLeNet + LSTM · HMDB-51 | average classification accuracy: 43.1 -> 49.3 | Helped |
| add class-specific 3D-skeleton feature constraints (ablation) | GoogLeNet + LSTM · UCF101 | average classification accuracy: 75.2 -> 81.5 | Helped |
| add class-specific 3D-skeleton feature constraints (ablation) | GoogLeNet + LSTM · HMDB-51 | average classification accuracy: 43.1 -> 51.4 | Helped |
| combine class-independent and class-specific skeleton constraints (ablation) | GoogLeNet + LSTM · UCF101 | average classification accuracy: 75.2 -> 86.9 | Helped |
| combine class-independent and class-specific skeleton constraints (ablation) | GoogLeNet + LSTM · HMDB-51 | average classification accuracy: 43.1 -> 55.3 | Helped |
| use class-specific rather than class-independent skeleton constraints (ablation) | GoogLeNet + LSTM · Sports-1M | Hit@1: 73.4 -> 62.2 | Hurt |
| use class-specific rather than class-independent skeleton constraints (ablation) | GoogLeNet + LSTM · UCF101 | average classification accuracy: 78.3 -> 81.5 | Helped |
| use hybrid steepest descent rather than Adam for regularized training (ablation) | RLSTM-g3 · UCF101 | average classification accuracy: 81.8 -> 86.9 | Helped |
| use hybrid steepest descent rather than Adam for regularized training (ablation) | RLSTM-g3 · HMDB-51 | average classification accuracy: 51.3 -> 55.3 | Helped |
| use hybrid steepest descent rather than AdaGrad for regularized training (ablation) | RLSTM-g3 · UCF101 | average classification accuracy: 80.7 -> 86.9 | Helped |
| use hybrid steepest descent rather than AdaGrad for regularized training (ablation) | RLSTM-g3 · HMDB-51 | average classification accuracy: 48.6 -> 55.3 | Helped |
Closest papers by meaning
- Co-Occurrence Feature Learning for Skeleton Based Action Recognition Using Regularized Deep LSTM Networks (AAAI 2016 · similarity 0.92)
- A New Representation of Skeleton Sequences for 3D Action Recognition (CVPR 2017 · similarity 0.90)
- An Attention Enhanced Graph Convolutional LSTM Network for Skeleton-Based Action Recognition (CVPR 2019 · similarity 0.90)
- An End-to-End Spatio-Temporal Attention Model for Human Action Recognition from Skeleton Data (AAAI 2017 · similarity 0.90)
- Unsupervised Representation Learning With Long-Term Dynamics for Skeleton Based Action Recognition (AAAI 2018 · similarity 0.89)
- Heterogeneous Skeleton-Based Action Representation Learning (CVPR 2025 · similarity 0.86)
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.