Researchers from the Autonomous University of Barcelona (UAB) have developed a newcomputer tool that, among other things,helps computers determine if a face falls into the category of attractive face, according to a study published in the journalPLoS ONE.
The software created by the team led by Mario Rojas, from the UAB Center for Computer Vision, in collaboration with researchers from the Department of Psychology at the University of Princenton (USA), is capable of predicting nine judgments based on facial features:attractive, competent, reliable, dominant, tight-fisted, fearful, outgoing, threatening, and likeable. To achieve this they have used machine learning techniques, a branch of artificial intelligence that uses examples to teach a program how to work. Three traits – dominant, threatening, and stingy – were predictable with an effectiveness ranging from 91 to 96 percent.
Among other things, the study aimed to discover what kind of information is, at a computer level, useful for the prediction task. For example, they found thatthe area around the eyes contains more information about attractiveness, whilethe area around the mouth presents more information about extroversion. The researchers also challenged the ability of their predictive program in front of the faces of various celebrities and found that their results were very consistent with the general ideas that people have about these public figures.
Facial features play a key role in our daily judgments about others. If the information on which the face evaluation is based could be learned automatically, it could be used as a tool to design better interactive systems, among other applications.
