Spring Break
The Daily OrangeComputer program evaluates physical beauty of women
By Elliott Townsend
Posted: 4/10/08, 10:47 PM EST Section: News
The old saying goes, "Beauty is in the eye of the beholder."
Now, the beholder is a computer.
For his master's thesis in computer science, Amit Kagian created a program that can "determine" a woman's facial attractiveness.
Under the guidance of professor Eytan Ruppin and professor Gideon Dror of Tel Aviv University, Kagian's research and exploration into programming and psychology has unearthed both the age-old controversy of beauty, as well as technological advancement.
In a series of stages, Kagian and his team polled a group of 30 participants. These participants were asked to judge roughly 100 pictures of Caucasian women on a scale of one to seven, based on the aesthetic beauty of the women in the pictures. In rating the pictures, the participants were instructed not to give reasons as to why they rated each woman the way they did.
After this first stage, Kagian processed the number ratings of each picture. After mathematically mapping each number, Kagian's research team came to a final set of 98 categories for characterizing a face, ranging from the shape of a woman's eyes to the width of her nose.
Kagian's team input the data into a computer along with new pictures of women who weren't rated in the first stage. The computer was then asked to rate each woman. The results surprised Kagian, as they appeared very similar to the results produced by humans.
Despite the similarities, the results are not universal, said Kagian in his final thesis paper.
Robin Riley, assistant professor in women's studies at Syracuse University, agrees with Kagain's assessment.
"These things are completely subjective," she said. "Each society has different cultural and gender preferences."
What Kagian's program arrived at was an "average beauty" based on the compilation of results from the human ratings. Regardless, Riley believes there are too many variables to take into account.
Like Riley, some feel that because beauty is based on an individual bias, it seems unlikely that a computer could process the same emotions.
Now, the beholder is a computer.
For his master's thesis in computer science, Amit Kagian created a program that can "determine" a woman's facial attractiveness.
Under the guidance of professor Eytan Ruppin and professor Gideon Dror of Tel Aviv University, Kagian's research and exploration into programming and psychology has unearthed both the age-old controversy of beauty, as well as technological advancement.
In a series of stages, Kagian and his team polled a group of 30 participants. These participants were asked to judge roughly 100 pictures of Caucasian women on a scale of one to seven, based on the aesthetic beauty of the women in the pictures. In rating the pictures, the participants were instructed not to give reasons as to why they rated each woman the way they did.
After this first stage, Kagian processed the number ratings of each picture. After mathematically mapping each number, Kagian's research team came to a final set of 98 categories for characterizing a face, ranging from the shape of a woman's eyes to the width of her nose.
Kagian's team input the data into a computer along with new pictures of women who weren't rated in the first stage. The computer was then asked to rate each woman. The results surprised Kagian, as they appeared very similar to the results produced by humans.
Despite the similarities, the results are not universal, said Kagian in his final thesis paper.
Robin Riley, assistant professor in women's studies at Syracuse University, agrees with Kagain's assessment.
"These things are completely subjective," she said. "Each society has different cultural and gender preferences."
What Kagian's program arrived at was an "average beauty" based on the compilation of results from the human ratings. Regardless, Riley believes there are too many variables to take into account.
Like Riley, some feel that because beauty is based on an individual bias, it seems unlikely that a computer could process the same emotions.
Be the first to comment on this story