A recognition approach for groups with interactions

Abstract

People often participate in activities in groups, such as buying goods in a shopping mall or walking around in a park. Interactive groups refer to the groups whose members have interactions such as shaking hands, embracing, which are not uncommon occurrences in our daily life. Existing group recognition approaches are based on the similarity of the individuals’ locations or signal features. The interactions among people are probably regarded as dissimilar and affect the recognition accuracy. Moreover, when not all group members perform the interactions, group recognition is even more difficult to achieve. In this paper, we propose an approach called Interactive Group Recognizing (IGR) for recognizing groups with interactions among their members. The actions of individuals are inferred based on the sensing data, and the disparity between two individuals is computed using the sliding window technique. After that, groups are recognized using a majority-voting based method. Experimental results show that compared with the existing approach, the average group recognition accuracy of IGR is improved by 6.9%.