Kinect AI breakthrough could reduce lag, improve accuracy

By Stephany Nunneley, Tuesday, 29 March 2011 22:23 GMT

Microsoft researchers have figured out a way to potentially reduce lag and improve accuracy in Kinect’s sensor.

Written by Microsoft Research in Cambridge, the research paper posted on the website I Programmer (thanks, GI.biz), touts new “AI learning routines,” which can be implemented to help Kinect identify and track different body parts.

The experiments held by the research division used millions of 3D depth maps which have already been used to identify body parts, and when combining the results, a 1000 core server could create an arrangement of data trees. These different data trees could then assign a body part to each piece of the 3D pixel data includes in the server.

Should Microsoft decide to implement the research, it could be easier for Kinect to predict joint placement in 3D while withdrawing the need for calibration. It would also considerably reduce lag when gaming.

If you enjoy tech reads, you should click the link and be happy.

Sometimes we include links to online retail stores. If you click on one and make a purchase we may receive a small commission. For more information, go here.

Comments