Research map / Video / Action recognition
Empowering Relational Network by Self-Attention Augmented Conditional Random Fields for Group Activity Recognition
Rizard Renanda Adhi Pramono, Yie Tarng Chen, Wen Hsien Fang
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 self-attention augmented CRF over max pooling (ablation) | Relational network · Volleyball group activity | accuracy: 87.1 -> 94.0 | Improved |
| add self-attention augmented CRF over max pooling (ablation) | Relational network · Collective Activity | accuracy: 84.6 -> 93.9 | Improved |
| add self-attention augmented CRF (ablation) | Relational network · Volleyball individual action | accuracy: 78.4 -> 82.5 | Improved |
| add unidirectional UTE (ablation) | CRF relational network · Volleyball group activity | accuracy: 94.0 -> 94.6 | Improved |
| replace unidirectional UTE with bidirectional UTE (ablation) | CRF relational network · Volleyball group activity | accuracy: 94.6 -> 95.0 | Improved |
| replace unidirectional UTE with bidirectional UTE (ablation) | CRF relational network · Collective Activity | accuracy: 94.8 -> 95.2 | Improved |
| add spatial location features (ablation) | Relational network · Volleyball group activity | accuracy: 94.1 -> 94.5 | Improved |
| add spatial location features (ablation) | Relational network · Collective Activity | accuracy: 93.9 -> 94.6 | Improved |
| add pose features (ablation) | Relational network · Volleyball group activity | accuracy: 94.5 -> 95.0 | Improved |
| add pose features (ablation) | Relational network · Volleyball individual action | accuracy: 82.3 -> 83.1 | Improved |
| add pose features (ablation) | Relational network · Collective Activity | accuracy: 94.6 -> 95.2 | Improved |
| use adaptive iterations instead of fixed 1 iteration (ablation) | CRF · Volleyball group activity | accuracy: 92.1 -> 95.0 | Improved |
Closest papers by meaning
- Hierarchical Relational Networks for Group Activity Recognition and Retrieval (ECCV 2018 · similarity 0.90)
- Learning Actor Relation Graphs for Group Activity Recognition (CVPR 2019 · similarity 0.90)
- Convolutional Relational Machine for Group Activity Recognition (CVPR 2019 · similarity 0.90)
- GroupFormer: Group Activity Recognition With Clustered Spatial-Temporal Transformer (ICCV 2021 · similarity 0.89)
- Detector-Free Weakly Supervised Group Activity Recognition (CVPR 2022 · similarity 0.89)
- stagNet: An Attentive Semantic RNN for Group Activity Recognition (ECCV 2018 · similarity 0.88)
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.
