Research map / Video / Action recognition
A Hierarchical Deep Temporal Model for Group Activity 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 |
|---|---|---|---|
| Use the two-stage hierarchical model instead of the strongest listed deep baseline | Group activity recognition · Collective Activity Dataset | Accuracy: 80.6 -> 81.5 | Helped |
| Use the two-stage hierarchical model instead of the cardinality kernel | Group activity recognition · Collective Activity Dataset | Accuracy: 83.4 -> 81.5 | Hurt |
| Use the two-stage hierarchical model instead of the strongest baseline | Team activity recognition · Volleyball dataset | Accuracy: 49.7 -> 51.1 | Helped |
| Add group-level LSTM 2 (ablation) | Two-stage hierarchical model · Collective Activity Dataset | Accuracy: 76.8 -> 81.5 | Helped |
| Add person-level LSTM 1 (ablation) | Two-stage hierarchical model · Collective Activity Dataset | Accuracy: 70.1 -> 81.5 | Helped |
| Add group-level LSTM 2 (ablation) | Two-stage hierarchical model · Volleyball dataset | Accuracy: 49.7 -> 51.1 | Helped |
| Add person-level LSTM 1 (ablation) | Two-stage hierarchical model · Volleyball dataset | Accuracy: 48.8 -> 51.1 | Helped |
| Replace max pooling with sum or average pooling (ablation) | Two-stage hierarchical model · Both datasets | Accuracy | Hurt |
| Add temporal modeling with LSTMs to frame-based baselines | Collective Activity Dataset group activity recognition · Collective Activity Dataset | Accuracy: 63.0 -> 64.2; 61.8 -> 62.2; 66.3 -> 81.5 | Mixed |
| Use whole-image temporal features instead of single-frame image classification | Group activity recognition · Volleyball dataset | Accuracy: 46.7 -> 37.4 | Hurt |
Closest papers by meaning
- Recurrent Modeling of Interaction Context for Collective Activity Recognition (CVPR 2017 · similarity 0.91)
- Convolutional Relational Machine for Group Activity Recognition (CVPR 2019 · similarity 0.89)
- CERN: Confidence-Energy Recurrent Network for Group Activity Recognition (CVPR 2017 · similarity 0.89)
- Detector-Free Weakly Supervised Group Activity Recognition (CVPR 2022 · similarity 0.89)
- Bi-Causal: Group Activity Recognition via Bidirectional Causality (CVPR 2024 · similarity 0.85)
- Learning Group Activity Features Through Person Attribute Prediction (CVPR 2024 · 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.