A U.K. Ph.D. student has developed a smart camera that can not only follow objects, but can learn from its tracking mistakes and then recalculate to track more accurately.
Zdenek Kalal from the University of Surrey's Centre for Vision, Speech and Signal Processing, developed the Predator system to help the disabled use computers. For example, Predator can "track a tip of a pen held in the mouth and act as a computer mouse for a paralyzed person."
But a system enabling motion tracking setups to learn from mistakes and improve their function could mean upgrades to everything from consumer trackpads to weapons systems.
The Predator system resembles already established motion-controlled technology such as Xbox 360's Kinect, appliance-friendly SoftKinetic gesture recognition systems, and emerging models of motion-capturing Web cameras like the CP Technologies Deluxe.
What sets it apart is its ability to learn from its errors and adjust its behavior--just as a human would gain better dexterity with experience. … Read more