TAU student develops software that ranks facial attractiveness
Key achievement is that the computer operates according to perceptions of beauty that were not input into it.
Most people can tell you if the person they are looking at is attractive, but they can't tell you why they think so. Now, a Tel Aviv University student has developed software to crack the age-old problem of identifying facial features that would be considered beautiful by most people.
"Until now, computers have been taught to identify basic facial characteristics - like is this a woman's face or a man's," explains Amit Kagian, who developed the program for his master's thesis in computer science.
"Our software allows the computer to complete a much more complex task of esthetic judgment, which humans cannot define exactly how they do it. Esthetic judgment is linked to sentiment and more abstract considerations, but now we have made the computer do it. This constitutes a substantial advance in the development of artificial intelligence."
The research, a combination of computer programming and psychological research, was conducted under professors Eytan Ruppin and Gideon Dror was recently published by the scientific journal Vision Research.
In the first stage, 30 human participants were asked to rate from 1-7 the beauty of several dozen pictures. Participants did not say why they ranked certain faces as more beautiful than others. The pictures were then processed and mathematically mapped.
"We came up with 98 numbers that represent the geometric shape of the face, as well as characteristics like hair color, smoothness of skin and facial symmetry," Kagian explains. Participants' rankings of the pictures were also input in the computer.
Now the computer's prediction capability was then tested. "We input new pictures of faces into the computer and it graded them based on the information it had." Human subjects were then asked to rank the new pictures too.
"The computer produced impressive results: the rankings were very similar to the rankings people gave."
According to Kagian, the key achievement is that the computer operated according to certain perceptions of beauty that were not input into it, but learned by processing the data it received. "These are positions people are unaware of and were not explicitly programmed into the computer. But the machine operated according to the positions."
The research revealed that faces considered beautiful are average - with no extreme facial characteristics. "The computer learned a mathematical function, however it implicitly learned to prefer average faces," Kagian says. He explained that although people have different opinions of beauty, a large enough sample group will identify a high level of agreement even involving subjects from different cultures.
Nonetheless, the experiment only involved women's faces, as there is a greater variety of positions regarding male beauty.