Research map / Video / Action recognition
Joint Learning of Social Groups, Individuals Action and Sub-group Activities in Videos
Mahsa Ehsanpour, Alireza Abedin, Fatemeh Saleh, Javen Shi, Ian Reid, Hamid Rezatofighi
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 individual action loss (joint training) (ablation) | group-only I3D · Volleyball | group accuracy: 91.0 -> 92.0 | Improved |
| add individual action loss (joint training) (ablation) | group-only I3D · Collective Activity | group accuracy: 84.6 -> 88.2 | Improved |
| add self-attention module (ablation) | Ours[w/o GAT] · Volleyball | individual action accuracy: 82.0 -> 83.2 | Improved |
| add self-attention module (ablation) | Ours[w/o GAT] · Collective Activity | individual action accuracy: 73.4 -> 75.3 | Improved |
| add self-attention module (ablation) | Ours[w/o GAT] · Volleyball | group accuracy: 92.0 -> 92.5 | Improved |
| add graph attention (GAT) module (ablation) | Ours[final] · Volleyball | group accuracy: 92.5 -> 93.1 | Improved |
| add graph attention (GAT) module (ablation) | Ours[final] · Collective Activity | group accuracy: 88.3 -> 89.4 | Improved |
| add graph attention (GAT) module (ablation) | Ours[final] · Collective Activity | individual action accuracy: 75.3 -> 78.3 | Improved |
| add graph attention (GAT) module (ablation) | Ours[final] · Volleyball | individual action accuracy: 83.2 -> 83.3 | No clear effect |
| add graph edge loss to learn social groups (ablation) | Ours[cluster] · Collective Activity social task | membership accuracy: 78.2 -> 83.0 | Improved |
| add graph edge loss to learn social groups (ablation) | Ours[cluster] · Collective Activity social task | social activity accuracy: 52.2 -> 69.0 | Improved |
| add graph edge loss to learn social groups (ablation) | Ours[cluster] · Collective Activity social task | individual action accuracy: 79.5 -> 83.3 | Improved |
Closest papers by meaning
- Hunting Group Clues with Transformers for Social Group Activity Recognition (ECCV 2022 · similarity 0.92)
- Social Scene Understanding: End-To-End Multi-Person Action Localization and Collective Activity Recognition (CVPR 2017 · similarity 0.91)
- GroupFormer: Group Activity Recognition With Clustered Spatial-Temporal Transformer (ICCV 2021 · similarity 0.91)
- JRDB-Act: A Large-Scale Dataset for Spatio-Temporal Action, Social Group and Activity Detection (CVPR 2022 · similarity 0.90)
- Detector-Free Weakly Supervised Group Activity Recognition (CVPR 2022 · similarity 0.89)
- Self-Supervised Social Relation Representation for Human Group Detection (ECCV 2022 · similarity 0.89)
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.
