Research map / Label-efficient and robust learning / Recognition and long-tail
Beyond F-Formations: Determining Social Involvement in Free Standing Conversing Groups From Static Images
Lu Zhang, Hayley Hung
Check 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 |
|---|---|---|---|
| Learn location-dependent frustum parameters instead of using global parameters | F-formation detection · Idiap Poster Data, T=1 | F1: 0.83 -> 0.94 | Improved |
| Learn location-dependent frustum parameters instead of using global parameters | F-formation detection · Idiap Poster Data, T=2/3 | F1: 0.94 vs 0.94 | No clear effect |
| Model associates with social-prior features | Associate detection · Idiap Poster Data | F1: 0.44 -> 0.71 | Improved |
| Use associate detections to remove false-positive F-formation members | F-formation detection · Idiap Poster Data, full-agreement formations | F1: 0.86 -> 0.91 | Improved |
| Add associate-detection feedback to global-F | F-formation detection · Idiap Poster Data, full-agreement formations | F1: 0.83 -> 0.88 | Improved |
| Use proximity and orientation features without group-size features (ablation) | social-A · Idiap Poster Data, associate detection | F1: 0.71 -> 0.70 | No clear effect |
| Use proximity and group-size features without orientation features (ablation) | social-A · Idiap Poster Data, associate detection | F1: 0.71 -> 0.67 | Hurt |
| Use orientation and group-size features without proximity features (ablation) | social-A · Idiap Poster Data, associate detection | F1: 0.71 -> 0.69 | No clear effect |
| Use proximity features only (ablation) | social-A · Idiap Poster Data, associate detection | F1: 0.71 -> 0.69 | Hurt |
| Use orientation features only (ablation) | social-A · Idiap Poster Data, associate detection | F1: 0.71 -> 0.69 | Hurt |
| Use group-size features only (ablation) | social-A · Idiap Poster Data, associate detection | F1: 0.71 -> 0.63 | Hurt |
| Extract social-prior features with global-F rather than spatial-F (ablation) | Associate detection · Idiap Poster Data | F1: 0.71 vs 0.71 | No clear effect |
Closest papers by meaning
- Part-Aware Bottom-Up Group Reasoning for Fine-Grained Social Interaction Detection (NeurIPS 2025 · similarity 0.84)
- The Audio-Visual Conversational Graph: From an Egocentric-Exocentric Perspective (CVPR 2024 · similarity 0.84)
- Modeling Multimodal Social Interactions: New Challenges and Baselines with Densely Aligned Representations (CVPR 2024 · similarity 0.84)
- MTGS: A Novel Framework for Multi-Person Temporal Gaze Following and Social Gaze Prediction (NeurIPS 2024 · similarity 0.84)
- Generative Proxemics: A Prior for 3D Social Interaction from Images (CVPR 2024 · similarity 0.84)
- Spot The Ball: A Benchmark for Visual Social Inference (CVPR 2026 · 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.
