MUMBAI: multi-person, multimodal board game affect and interaction analysis dataset

Abstract Board games are fertile grounds for the display of social signals, and they provide insights into psychological indicators in multi-person interactions. In this work, we introduce a new dataset collected from four-player board game sessions, recorded via multiple cameras, containing over 46 hours visual material. The MUMBAI is extensively annotated with emotional moments all sessions. Additional data comes personality experience questionnaires. Our four-person setup allows investigation non-verbal interactions beyond dyadic settings. We present three benchmarks expression detection emotion classification discuss potential research questions analysis group dynamics during games.