You Can Give the Kinect the Power of Image Recognition
The Kinect can see, but the ability to see objects is different from the ability to recognize objects. You and...
The Kinect can see, but the ability to see objects is different from the ability to recognize objects. You and I, with our eyes and brains that work so effectively, can see a water bottle of pretty much any size, shape, color, or material, and recognize what it’s for. But a Kinect is not as smart as we are, and needs a hand to get to our level. That’s where you come in.
A team of Swedish researchers is asking Kinect users to use their Kinects as a 3-D scanner to take pictures of various objects for a project known as Kinect@Home. All of those scans will go into a library all Kinects will be able to access, allowing the device to recognize all kinds of different objects. It’s obviously valuable; pretty much everything we’d like the Kinect to do could be enhanced by giving it the ability to distinguish between and identify various materials. Only a few days in, Kinect@Home already has nearly 150 scans.
The project is designed for household robots–imagine mounting a Kinect on some kind of wheeled creation, and not only can it see the shapes of objects in a room–it can recognize what they are. “Don’t go near the vase!” you could say. And it could understand you! Pretty amazing stuff.