A paper by Microsoft researchers details potential improvements to the accuracy and speed of the Kinect’s sensing processes, though we still don’t know if the company will use these findings.
The paper details new AI learning routines that can be used to improve Kinect’s ability to see and track parts of the body.GIsays this paper, written by Microsoft Research in Cambridge, used the millions of 3D depth maps already identified with body parts. A server pulled together all the data from these maps. This data could be used to predict joint placement more quickly in 3D space. This could remove the need for calibration as well as improve the lag times of Kinect.
I don’t know how easy it would be to implement this into Kinect games. Let’s hope that Microsoft will use it if they can.
