Multi-model hypothesis tracking of groups of people in RGB-D data

Detecting and tracking people and groups of people is a key skill for intelligent vehicles, interactive systems and robots that are deployed in humans environments. In this paper, we address the problem of detecting groups of people from learned social relations between individuals with the goal to reliably track group formation processes. Opposed to related work, we track and reason about multiple social grouping hypotheses in a recursive way, assume a mobile sensor that perceives the scene from a first-person perspective, and achieve good tracking performance in realtime using RGB-D data. In experiments in large-scale outdoor data sets, we demonstrate how the approach is able to track groups of people with varying sizes over long distances with few track identifier switches.