Voice commands are de rigueur these days with products like Siri and Google Glass, but what about people who can’t speak? Sign language is the easiest way for many deaf people to communicate, but it has been difficult for computers to readily understand this complex form of communication until recently. Now, researchers in China are using Microsoft’s Kinect for Windows SDK to actively translate sign language into written text.
Microsoft Research Asia and the Institute of Computing Technology at the Chinese Academy of Sciences have teamed up to use Kinect for Windows to effectively track complex hand motions in 3D space. By combining the data from both the RGB camera and the depth-sensing infrared camera in the Kinect, these researchers were able to develop an impressive system to aid communication between the deaf and the hearing.
The way this system is designed to work is actually quite clever — novel even. By recording and then normalizing the movements of sign language, the system uses an algorithm to determine the “3D motion-trajectory alignment.” Once the computer crunches the visual data, it can then match it with known words; ranked by relevance. This effectively allows the Kinect to serve as a translation engine that can enable complex communication between the hearing and the deaf.
Most impressively, this works with inexpensive off-the-shelf technology instead of costly proprietary systems. Researchers all over the world are also using the low-cost Kinect for applications ranging from diagnosing depression to allowing for the development of precognitive robots. Considering how well these projects work with existing technology, just imagine the projects that will be possible when the next-gen Kinect starts shipping.
PrimeSense, the company that developed the technology used in the first-gen Kinect, is now rumored to be in Apple’s crosshairs. Apple’s interest in the Israeli company is well documented, and an outright purchase could bring some truly valuable technology directly to Apple’s products. With PrimeSense’s motion, depth, and voice-tracking tech in its pocket, Apple could easily jump to the front of the pack in terms of human-computer interfacing. With any luck, this news will drive Microsoft to work even harder with its Kinect research and development. The more competition we have in this emerging market, the better off we’ll all be.
Research paper: Sign Language Recognition and Translation with Kinect (PDF)
[Image credit: Valerie Everett]
Source Article from http://www.extremetech.com/extreme/161771-chinese-researchers-use-the-kinect-to-translate-sign-language-to-text
Chinese researchers use the Kinect to translate sign language to text