UCF researchers designing new facial recognition technology
By Stuart Campbell, Web Producer -
Who's your daddy? A team from the University of Central Florida has designed a computer program to answer just that question.
Graduate student Afshin Dehfghan is leading the research team responsible for developing a facial recognition tool able to quickly match pictures of children with their biological parents. They started the project with more than 10,000 online images of celebrities, politicians and their children.
"The machine was asked to rank the images based on the similarity to the given image, so as you can see we input a George Clooney image and his right biological father was ranked first in our system," reveals Afshin.
While humans might look for something like Catherine Zeta Jones' smile in her offspring, the computer is able to map out and match a digital checkerboard of facial features like the shape of an eye, a chin or distinctive parts of a forehead.
"When we let the computer pick those parts automatically, we were able to get an even better performance," admits Afshin.
In layman's terms, they're teaching a computer to think after feeding it all those images and creating a memory bank with the computer program. It's an age old question still being debated, can computers think? Dehgan and his team believe they are one step closer to answering that question.
"It can learn pretty well. It can do a lot better than chance and hopefully this can be of benefit to many applications, like homeland security and even the police department using this software,"says Afshin.