Research map / Detection and segmentation / CNN features
Socially Constrained Structural Learning for Groups Detection in Crowd
Francesco Solera, Simone Calderara, Rita Cucchiara
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 |
|---|---|---|---|
| learn feature weights with structural SVM instead of random equal weights (ablation) | group detection · hotel | G-MITRE precision: 71.0 -> 97.3 | Improved |
| learn feature weights with structural SVM instead of random equal weights (ablation) | group detection · student003 | G-MITRE precision: 59.9 -> 81.7 | Improved |
| learn feature weights with structural SVM instead of random equal weights (ablation) | group detection · eth | G-MITRE recall: 65.2 -> 94.2 | Improved |
| train with G-MITRE loss instead of pairwise loss (ablation) | structural SVM group detection · stu003 | G-MITRE precision: 68.9 -> 80.1 | Improved |
| train with G-MITRE loss instead of pairwise loss (ablation) | structural SVM group detection · eth | G-MITRE precision: 88.7 -> 91.9 | Improved |
| train with G-MITRE loss instead of pairwise loss (ablation) | structural SVM group detection · hotel | G-MITRE precision: 90.1 -> 89.2 | No clear effect |
| use learned method instead of Ge et al. clustering | group detection · hotel | G-MITRE precision: 89.2 -> 97.3 | Improved |
| use learned method instead of Ge et al. clustering | group detection · student003 | G-MITRE precision: 77.2 -> 81.7 | Improved |
| replace ground-truth trajectories with detector and tracker output | proposed method · hotel | G-MITRE precision: 97.3 -> 77.9 | Hurt |
| replace ground-truth trajectories with detector and tracker output | proposed method · student | G-MITRE precision: 81.7 -> 75.0 | Hurt |
| use proposed method instead of Yamaguchi et al. on automatic trajectories | group detection · hotel | G-MITRE precision: 46.3 -> 77.9 | Improved |
| use sequential or online training instead of batch (ablation) | proposed method · gal1 of GVEII with growing density | F1 | Improved |
Closest papers by meaning
- A Multiview-Based Parameter Free Framework for Group Detection (AAAI 2017 · similarity 0.87)
- Quantifying and Detecting Collective Motion by Manifold Learning (AAAI 2017 · similarity 0.87)
- Detecting Coherent Groups in Crowd Scenes by Multiview Clustering (TPAMI 2018 · similarity 0.86)
- Self-Supervised Social Relation Representation for Human Group Detection (ECCV 2022 · similarity 0.86)
- Part-Aware Bottom-Up Group Reasoning for Fine-Grained Social Interaction Detection (NeurIPS 2025 · similarity 0.83)
- Social Physics Informed Diffusion Model for Crowd Simulation (AAAI 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.
