Researchers at Georgia Tech are working to move Kinect out of the toy category and move it into that of “useful device.”
The video below is a presentation of the ongoing effort in American Sign Language recognition using the Kinect by researchers at the College of Computing, Georgia Institute of Technology.
Originally achieved by pairing Kinect hardware with a set of knitted gloves that contained accelerometers, the researchers have now managed to ditch the gloves and focus directly on movements. By measuring the distances and changes between various body parts, the group has managed to achieve results of no lower than 98.8% accuracy based on tests of increasing difficulty.
The researchers are now working to integrate “hand shape features” rather than just gestural movements, so that they can expand the vocabulary and create a useful ASL tool. The focus of their development is the CopyCat software, which’s to teach deaf children how to communicate with ASL.
The required hand shape features may need a higher resolution image than is currently provided, but it’s rumoured that Microsoft would only have to push out a firmware update for this to be possible.
This’s only the first of many possible medical, academic and accessibility uses of the gaming device.