Astonishing age progression software may be important tool

SEATTLE -- A potentially important new tool to find missing children was unveiled at the University of Washington this week. Researchers used "big data" to create astonishingly accurate age progression software.

The reason this new software is so accurate is that it's based on 40,000 images of random people of all ages grabbed from the internet.

It allows Ira Kemelmacher-Shlizerman and her team at the UW Paul Allen Computer Science Center to use a child's simple photo and show what he or she would look like through the years to 80 and older.

Their software doesn't make you look like an average older person or the aggregate of all those thousands of random photos. It makes your nose, your eyes, and your mouth look older. It stays true to you. There is no artistic interpretation that a sketch artist or photo shop artist must use because it's based solely on the power of big data.

For example, a little girl's distinct eyes can also be seen in the old age photo. It stays true to ethnicity. And it even retains charming smiles.

When the computer images are put next to the actual photos of the real people, it's hard to tell which photos are real.

When Kemelmacher-Shlizerman and her team showed special comparative photos through surveys to thousands of people around the world, the software scored well.

"People couldn't distinguish between our result and the real photo," she said.

The obvious potential use is to help find missing kids. Kemelmacher-Shlizerman contacted the Center for Missing and Exploited Children and was told her software's ability to utilize even a single toddler photo can be valuable because "photos of kids from zero to five years old are the most difficult ones to produce even for an artist. And they are actually the most important ones," she said.

The software could improve existing age progression in the movies. It could be used for cosmetic reasons. Or it could be used for pure entertainment at parties or online.

But all that is the future. Kemelmacher-Shlizerman implied that perhaps the most important application in many ways would be for the software developed by her team to help find a missing child.