Knowledge Bank

The advent of modern sensing technologies has allowed data collection on people and events to occur at an unprecedented scale. One such application has been to study different facets of early childhood behavior. Research has shown that when children first start compulsory education, their environment and interactions with peers and figures of authority can either boost their development, or become a basis for negative outcomes. Specifically, peer-to-peer isolation can occur as early as 3-5 years of age, and can be especially devastating to developing children. This thesis outlines the process of designing an experiment to collect video data from a preschool classroom, and develop scientific, quantitative methods for studying isolation, using computer vision techniques. While this phenomenon is easy to understand, it is challenging to rigorously define and identify. Preliminary results here show that an "isolation score" that is based on computer vision data is a promising method for identifying child isolation in a classroom