The MatchNMingle Dataset: A Novel Multi-Sensor Resource for the Analysis of Social Interactions and Group Dynamics In-the-Wild During Free-Standing Conversations and Speed Dates

We present MatchNMingle, a novel multimodal/multisensor dataset for the analysis of free-standing conversational groups and speed-dates in-the-wild. MatchNMingle leverages use wearable devices overhead cameras to record social interactions 92 people during real-life speed-dates, followed by cocktail party. To our knowledge, has largest number participants, longest recording time set manual annotations actions available in this context scenario. It consists 2 hours data from acceleration, binary proximity, video, audio, personality surveys, frontal pictures speed-date responses. Participants' positions group formations were manually annotated; as (eg. speaking, hand gesture) 30 minutes at 20 FPS making it first incorporate annotation such cues context. an empirical performance crowdsourcing workers against trained annotators simple complex tasks, founding that although efficient using more tasks like action led additional poor inter-annotator agreement compared (differences up 0.4 Fleiss' Kappa coefficients). also provide example experiments how can be used.