â€śYou are going to be so handsome/pretty when you grow up!â€ť
We hear that all the time when people talk about their babies, but I have always been one who looks at an infant and wonders, â€śHow in the world can you tell?â€ť Sure, a measure of that is just the parent or family member dotting on the child, but seriously how can anyone tell how attractive someone is going to be as an adult when they are only a baby? I mean, sure this is not one of those great mysteries of the universe, or even something I can admit giving a great deal of thought to, but when it comes up it always leaves me curious.
Finally, we will have a much more definitive way of determining how right these people are thanks to new software developed by researchers at the University of Washington who have created a program that will take a picture of a baby and age that picture fairly accurately all the way up through the ages of 80-100. This is the first fully automated system for aging babies to adults that works with variable lighting, poses, and expressions. This software does this by using the average of thousands of faces of the same age and sex, then calculates the changes between groups as they age to make changes to the given infant’s face. It also determines the average pixel arrangement using thousands of random Internet pictures of faces in different age and gender brackets. It then finds similarities between these averages from each bracket and then uses that to calculate an average change in facial shape and appearance between the different ages. All of these changes are then applied to the infant’s picture to predict just how they are are likely to appear for any subsequent age up to 80-100.
In order to test how well this software actually works, the researchers tested their rendered images against 82 actual people photographed over the course of their lifetimes. For their experiment, they asked random users to identify which was the rendered photo and which was the actual one for each example. What they found were that users chose the rendered picture nearly as often as the real-life pictures, as they could tell little difference between them. According to Steven Seitz, a professor of computer science and engineering at the University of Washington, â€śOur extensive user studies demonstrated age progression results that are so convincing that people can’t distinguish them from reality. When shown images of an age-progressed child photo and a photo of the same person as an adult, people are unable to reliably identify which one is the real photo.â€ť
Of course, the software is not yet perfect. While it does consider gender and age, the team hopes to incorporate other factors that it will take into account, such as ethnicity and cosmetic factors like hair whitening and wrinkles that way they can create a robust enough technique to represent every possible human face. While this technology will indeed help in answering the age-old question of â€śWhat will my baby look like when they are older,â€ť the greater application for it will be in rendering older images of missing children. Today, this process is often done by hand and takes a lot of time, effort, and guesswork. This software can do the same work in only a fraction of the time, and with a much higher potential degree of accuracy, which could potentially lead to more of these tragic cases finding its own resolution.
Image Credit: Thinkstock