The University of Illinois said that it's developing software which tells how old an individual is by looking at her or his face.

According to Thomas S. Huang, a professor of electrical and computer engineering at the University, the software will have applications which include IDing underage drinkers, stop kids from buying cigarettes, and stop children from getting onto pornographic web sites.

The age estimation software has three components - face detection, discrimination, manifold learning and multiple linear regression and was trained on a database with photographs of 1,600 faces. It can estimate ages from one to 93 years, but the accuracy is only 50% within five years of the subject, but 80% when estimating ages within 10 years.

Training can improve the accuracy of the software, said the university.  Other applications might include advertising gizmos which snap your face as you walk by a shop, and then display what might appeal to someone of a certain age. And it could also be used in CRM (customer relationship management) applications, to estimate the ages of people who buy Mars Bars, Lucozade, or Vimto. Or identify how old Oscar Wilde was, when he wrote The Picture of Dorian Gray. X