Research map / Video / Action recognition
Hierarchical Long Short-Term Concurrent Memory for Human Interaction Recognition
Xiangbo Shu, Jinhui Tang, Guo-Jun Qi, Wei Liu, Jian Jun Yang
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 temporal modeling with LSTM instead of per-frame CNN features only (ablation) | B2 vs B1 · BIT | accuracy (%): 71.88 -> 83.50 | Improved |
| model inter-related motion with Co-LSTSM instead of independent single-person LSTMs (B4) (ablation) | Co-LSTSM · BIT | accuracy (%): 88.50 -> 92.88 | Improved |
| add Single-Person LSTMs with hierarchical Co-LSTM instead of Co-LSTSM (ablation) | H-LSTCM · BIT | accuracy (%): 92.88 -> 94.03 | Improved |
| add Single-Person LSTMs with hierarchical Co-LSTM instead of Co-LSTSM (ablation) | H-LSTCM · UT | accuracy (%): 95 -> 98.33 | Improved |
| add hierarchical single-person dynamics and Co-LSTM instead of Co-LSTSM+ (ablation) | H-LSTCM · CAD | accuracy (%): 79.14 -> 83.75 | Improved |
| extended Co-LSTSM+ instead of multiple single-person LSTMs plus whole LSTM (B4) (ablation) | Co-LSTSM+ · CAD | accuracy (%): 78.46 -> 79.14 | No clear effect |
| add hierarchical Co-LSTM instead of B4 (ablation) | H-LSTCM · Volleyball | accuracy (%): 82.5 -> 88.4 | Improved |
| multiple LSTMs per person (B3) instead of merging persons into one box (B2) (ablation) | B3 vs B2 · CAD | accuracy (%): 63.71 -> 66.02 | Improved |
| hierarchical LSTM with single-person dynamics instead of best LSTM-based prior (Ke et al.) | H-LSTCM · BIT | accuracy (%): 85.20 -> 94.03 | Improved |
| H-LSTCM versus prior best on Volleyball (Shu et al.) | H-LSTCM · Volleyball | accuracy (%): 83.6 -> 88.4 | Improved |
| H-LSTCM versus Ibrahim et al. LSTM on CAD | H-LSTCM · CAD | accuracy (%): 81.50 -> 83.75 | Improved |
Closest papers by meaning
- Lattice Long Short-Term Memory for Human Action Recognition (ICCV 2017 · similarity 0.89)
- Recurrent Modeling of Interaction Context for Collective Activity Recognition (CVPR 2017 · similarity 0.88)
- A Hierarchical Deep Temporal Model for Group Activity Recognition (CVPR 2016 · similarity 0.88)
- Memory-Augmented Temporal Dynamic Learning for Action Recognition (AAAI 2019 · similarity 0.87)
- UCF-STAR: A Large Scale Still Image Dataset for Understanding Human Actions (AAAI 2020 · similarity 0.87)
- Temporal Dynamic Graph LSTM for Action-Driven Video Object Detection (ICCV 2017 · 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.
