Microsoft Research’s Advanced Technologies Lab in Israel is working on a research project named Project Prague that will allow app developers to include basic hand gesture support into their applications without much work. They can also create customized gestures for their own apps or other products, with very little additional programming or expertise.
Let’s say you’re talking to a colleague over Skype and you’re ready to end the call. What if, instead of using your mouse or keyboard to click a button, you could simply make the movement of hanging up the phone?
Need to lock your computer screen quickly? What if, instead of scrambling to close windows and hit keyboard shortcuts, you simply reach out and mimic the gesture of turning a key in a lock?
Project Prague utilizes machine learning to train systems to recognize motions, runs using a retail 3D camera. MSR recorded millions of hand images and then used that data set to train the technology to recognize every possible hand pose and motion.
Eyal Krupka, a principal applied researcher and head of the lab’s computer vision and machine learning research, said the technology then uses hundreds of micro artificial intelligence units, each analyzing a single aspect of the user’s hand, to accurately interpret each gesture.
The end result is a system that doesn’t just recognize a person’s hand, but also understands that person’s intent.
Read more about this project here.