Research map / Video / Action recognition
Convolutional Relational Machine 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 |
|---|---|---|---|
| add a spatial activity map to feature-map input (ablation) | CRM on Volleyball · Volleyball | accuracy: 75.99 -> 89.82 | Helped |
| add a second refinement stage (ablation) | CRM on Volleyball · Volleyball | accuracy: 89.82 -> 90.72 | Helped |
| add refinement stages beyond stage 2 (ablation) | CRM on Volleyball · Volleyball | accuracy: 90.72; 90.42; 90.80 | No clear effect |
| add individual-activity fields to the activity map (ablation) | CRM on Volleyball · Volleyball | accuracy: 88.85 -> 90.80 | Helped |
| add learned global aggregation instead of direct box pooling (ablation) | CRM on Volleyball · Volleyball | accuracy: 87.80 -> 90.80 | Helped |
| combine feature map and activity map rather than use the activity map alone (ablation) | CRM on Volleyball · Volleyball | accuracy: 82.72 -> 90.80 | Helped |
| use I3D RGB video input instead of single-frame input (ablation) | CRM on Volleyball · Volleyball | accuracy: 90.80 -> 92.07 | Helped |
| use optical flow rather than RGB with I3D (ablation) | CRM on Volleyball · Volleyball | accuracy: 92.07 -> 91.47 | Hurt |
| fuse RGB and optical-flow predictions (ablation) | CRM on Volleyball · Volleyball | accuracy: 92.07; 91.47 -> 93.04 | Helped |
| use optical flow rather than RGB alone (ablation) | CRM on Collective Activity · Collective Activity | MCA: 83.41 -> 85.44 | Helped |
| use CRM rather than the strongest listed baseline | Collective Activity · Collective Activity | MCA: 87.20 -> 85.75 | Hurt |
| use CRM rather than the strongest listed baseline | Collective Activity with Walking and Crossing combined as Moving · Collective Activity | MPCA: 89.7 -> 94.2 | Helped |
Closest papers by meaning
- Learning Actor Relation Graphs for Group Activity Recognition (CVPR 2019 · similarity 0.91)
- A Hierarchical Deep Temporal Model for Group Activity Recognition (CVPR 2016 · similarity 0.89)
- Structure Inference Machines: Recurrent Neural Networks for Analyzing Relations in Group Activity Recognition (CVPR 2016 · similarity 0.88)
- Bi-Causal: Group Activity Recognition via Bidirectional Causality (CVPR 2024 · similarity 0.84)
- Learning Group Activity Features Through Person Attribute Prediction (CVPR 2024 · similarity 0.83)
- Learning from Synthetic Human Group Activities (CVPR 2024 · similarity 0.83)
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.