Research map / Federated learning and privacy / Privacy attacks and secure learning
ConfLab: A Data Collection Concept, Dataset, and Benchmark for Machine Analysis of Free-Standing Social Interactions in the Wild
Chirag Raman, Jose Vargas Quiros, Stephanie Tan, Ashraful Islam, Ekin Gedik, Hayley Hung
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 |
|---|---|---|---|
| Use acceleration instead of pose (MS-G3D) for speaking detection | speaking status detection · ConfLab | AUC: 0.676 -> 0.813 | Improved |
| Fuse MS-G3D pose with Minirocket acceleration (late fusion) | Minirocket · ConfLab speaking status | AUC: 0.813 -> 0.823 | Improved |
| Add camera 4 to training (double samples to 69k) (ablation) | Mask-RCNN keypoints · ConfLab camera 6 | performance: 260% | Improved |
| Add camera 8 (another 43k samples) (ablation) | Mask-RCNN keypoints · ConfLab camera 6 | performance: 46% | Improved |
| Increase training samples beyond 16% of frames (ablation) | Mask-RCNN · ConfLab camera 6 | performance: 16% | No clear effect |
| Use shoulder instead of head orientation (ablation) | GCFF · ConfLab F-formation | F1 (T=2/3): 0.47 -> 0.56 | Improved |
| Use shoulder instead of head orientation (ablation) | GTCG · ConfLab F-formation | F1 (T=2/3): 0.51 -> 0.46 | Hurt |
| Use hip instead of head orientation (ablation) | GTCG · ConfLab F-formation | F1 (T=2/3): 0.51 -> 0.45 | Hurt |
| Use hip instead of head orientation (ablation) | GCFF · ConfLab F-formation | F1 (T=1): 0.31 -> 0.25 | Hurt |
| Stricter tolerance threshold T=1 instead of T=2/3 (ablation) | GTCG with head orientation · ConfLab F-formation | F1: 0.51 -> 0.40 | Hurt |
| Stricter tolerance threshold T=1 instead of T=2/3 (ablation) | GCFF with shoulder orientation · ConfLab F-formation | F1: 0.56 -> 0.36 | Hurt |
Closest papers by meaning
- JRDB-Social: A Multifaceted Robotic Dataset for Understanding of Context and Dynamics of Human Interactions Within Social Groups (CVPR 2024 · similarity 0.83)
- Count What You Want: Exemplar Identification and Few-Shot Counting of Human Actions in the Wild (AAAI 2024 · similarity 0.83)
- MimeQA: Towards Socially-Intelligent Nonverbal Foundation Models (NeurIPS 2025 · similarity 0.83)
- WildChat-50M: A Deep Dive Into the Role of Synthetic Data in Post-Training (ICML 2025 · similarity 0.82)
- HOI-M^3: Capture Multiple Humans and Objects Interaction within Contextual Environment (CVPR 2024 · similarity 0.82)
- WildChat: 1M ChatGPT Interaction Logs in the Wild (ICLR 2024 · similarity 0.82)
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.
