Body-Part Tracking of Infants

Motion tracking is a widely used technique to analyze and measure adult human movement. However, these methods cannot be transferred directly to motion tracking of infants due to the big differences in the underlying human model. However, motion tracking of infants can be used for automatic analysis of infant development and might be able to tell something about possible motor disabilities such as cerebral palsy. In this paper, we address markerless 3D body part detection of infants using a widely available depth sensor and discuss some of the major challenges that arise. We present a method to detect and identify a set of the anatomical extremities and the results are evaluated based on manually annotated 3D positions.