Pattern recognition at its best. (via Bringham Young University)
Anyone that has seen a Sci-Fi movie can tell you that eventually robots will take over the world. Well, Bringham Young University engineer Dah-Jye Lee has officially made the first step towards devising a world where computers can think on their own.
Lee, Fellow BYU Electrical and Computing Engineer Professor James Archibald and graduate students Beau Tippetts and Kirt Lillywhite developed an algorithm that can accurately decipher the objects in various videos and pictures without human interference. How? The genetic algorithm is actually able to learn to identify objects on its own. Lee and his team based the algorithm off the process by which children learn to differentiate objects. Children learn to identify objects through pictures, so Lee and his team decided to try the same philosophy with the algorithm, and it worked. The computer receives a series of images and learns to decipher them on its own, with a 99-100% accuracy rate.
The computer is able to identify a range of objects, including species of fish, faces, airplanes, motorbikes and cars. The computer does not need to be reset before each identification – it seamlessly runs on its own.
Lee and his team think the algorithm may be most useful for the detection of invasive fish species in various bodies of water, but the possibilities are endless.
The study was published in December’s issue of Pattern Recognition. The team calls its development the “ECO features” genetic algorithm. There is no word yet on when the algorithm will be used in the real world, but we hope it’s a nice robot.
C
See more news at: