Research map / Probabilistic and scientific modeling / Dynamical systems
Structure Inference Machines: Recurrent Neural Networks for Analyzing Relations in 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 instance-level structure gates to message passing (ablation) | Structure Inference Machine · Collective Activity Dataset | accuracy: 74.33% -> 81.22% | Helped |
| add instance-level structure gates to message passing (ablation) | Structure Inference Machine · Collective Activity Extended Dataset | accuracy: 88.16% -> 90.23% | Helped |
| add instance-level structure gates to message passing (ablation) | Structure Inference Machine · Nursing Home Dataset | accuracy: 84.94% -> 85.50% | Helped |
| untie message-passing weights across iterations (ablation) | ungated Structure Inference Machine · Collective Activity Dataset | accuracy: 74.02% -> 74.33% | Helped |
| untie message-passing weights across iterations (ablation) | ungated Structure Inference Machine · Collective Activity Extended Dataset | accuracy: 87.97% -> 88.16% | Helped |
| untie message-passing weights across iterations (ablation) | ungated Structure Inference Machine · Nursing Home Dataset | accuracy: 84.91% -> 84.94% | Helped |
| untie message-passing weights across iterations (ablation) | gated Structure Inference Machine · Collective Activity Dataset | accuracy: 81.22% -> 81.22% | No clear effect |
| untie message-passing weights across iterations (ablation) | gated Structure Inference Machine · Collective Activity Extended Dataset | accuracy: 90.14% -> 90.23% | Helped |
| untie message-passing weights across iterations (ablation) | gated Structure Inference Machine · Nursing Home Dataset | accuracy: 85.32% -> 85.50% | Helped |
| increase inference iterations from 1 to 3 (ablation) | Gated Untied Weights · Collective Activity Dataset | accuracy: 80.12% -> 81.22% | Helped |
| increase inference iterations from 1 to 3 (ablation) | Gated Untied Weights · Collective Activity Extended Dataset | accuracy: 89.51% -> 90.23% | Helped |
| increase inference iterations from 1 to 2 (ablation) | Gated Untied Weights · Nursing Home Dataset | accuracy: 84.46% -> 85.50% | Helped |
Closest papers by meaning
- A Hierarchical Deep Temporal Model for Group Activity Recognition (CVPR 2016 · similarity 0.89)
- Learning Actor Relation Graphs for Group Activity Recognition (CVPR 2019 · similarity 0.88)
- Convolutional Relational Machine for Group Activity Recognition (CVPR 2019 · similarity 0.88)
- Learning Visual Context for Group Activity Recognition (AAAI 2021 · similarity 0.88)
- Progressive Relation Learning for Group Activity Recognition (CVPR 2020 · similarity 0.88)
- CERN: Confidence-Energy Recurrent Network for Group Activity Recognition (CVPR 2017 · similarity 0.87)
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.