Computer vision analysis for children's social play classification in peer-play scenarios

Abstract

Labeling children's social play behavior is an important process in children's peer-play analysis which is traditionally done by experienced coders. With the growing volume of data, automatic methods for labeling are increasingly required. This paper presents a novel method to classify children's social play behavior in peer-play scenarios into three categories (solitary play, parallel play and group play). Based on the two key cues attentiveness and proximity proposed in "The Play Observation Scale", unary features and pairwise features are calculated to describe the relationships between a child and the whole context, and the interactions between two children. Inspired by the recent studies in social behavior analysis and interaction recognition, children's activities are classified by support vector machine (SVM) and hidden conditional random field (HCRF). This method is evaluated by a dataset of children's peer-play scenarios collected by psychology researchers and the results show this method has a good performance in the dataset.