My Own Private Nightlife: Understanding Youth Personal Spaces from Crowdsourced Video

Private nightlife environments of young people are likely characterized by the particular ambiance, physical attributes, and activities but little is known about it. For instance, previous studies have documented ambiance and physical characteristics of homes using pictures from Foursquare or Airbnb, but there is a reasonable doubt that such staged data cannot reliably represent real-life situations. As a first attempt at describing the physical and ambiance features of homes using manual annotations and predicting ambiance characteristics using machine-extracted features, we used a unique dataset of 301 crowdsourced videos of home environments recorded in-situ by young people on weekend nights. Agreement among the five independent annotators was high for most features. Results of the annotation task revealed various patterns of youth home spaces features, such as the type of room attended (e.g., predominantly living room and bedroom), the number and gender of friends present, the type of ongoing activities (e.g., watching TV or computer alone; drinking, chatting and eating in the presence of others) and ambiance attributes with their correlations. Then, object and scene features of places, extracted by deep learning, were found to highly correlate with ambiances, while sound features mostly recognized ‘music’ and ‘speech’ only. Finally, the results of a regression task for predicting ambiances from those features showed that six of the ambiance categories can be inferred with R 2 in the [0.21, 0.69] range. Our work is quite novel with regard to the type of data (i.e., crowdsourced videos of real-life homes) and the analytical design (i.e., the combined use of manual annotation and deep learning to identify relevant features). This work potentially points to interesting ways to automatically predicting ambiances from videos of private environments at homes as a contribution to the multimedia community of researching ambiances at private residences.