Measuring attention using Microsoft Kinect
Darren Stanley · RIT Scholar Works (Rochester Institute of Technology) · 2013
The transfer of knowledge between individuals has increasingly become achieved with the aid of interfaces or computerized training applications. However, computer based training currently lacks the ability to monitor human behavioral changes and respond to them accordingly. This study examines the ability to predict user attention using features of body posture and head pose. Predictive abilities are assessed by an analysis of the relationship between the measured posture features and common objective measures of attention, such as reaction time and reaction time variance. Subjects were asked to participate in a series of sustained attention tasks while aspects of body movement and positioning were recorded using a Microsoft Kinect. Results showed support for identifiable patterns of behavior associated with attention while also suggesting the complex inter-relationship of measured features and susceptibility of these features to environmental conditions.