Imagine asking, “Mirror, mirror on the wall, who’s the fairest of them all?” and having the mirror actually answer. Someday soon it may — if the mirror contains computer chips programmed to run new software developed by Canadian university undergraduate student Joshua Chauvin.
Chauvin and fellow Canadian researches have found that a computer can be trained to rate human facial attractiveness the same way that people do.
To test their creation, the researchers asked a group of people to rate the attractiveness of other people in images, and then asked the computer to follow suit. Eighty-five percent of the time the computer’s ratings fell within one point of the human rating.
According to Chauvin…
“So what (the neural network is) doing is coming up with its own rating for those images based on some sort of understanding of what it thinks the population finds as attractive.”
The technology isn’t ready to start replacing beauty pageant judges yet, but who knows? To his credit, Chauvin sees the potential for more practical uses for it, like one day using the software to corroborate a doctor’s diagnoses by identifying certain facial characteristics of some diseases.
Perhaps less altruistically, he also sees the software being used by marketing and advertising companies to assess how their target audience may respond to the models hired for different ad campaigns.
This young researcher’s name (Chauvin) sounded French to me, so I pulled up Yahoo! Babel Fish to check the translation.
What are the odds? 🙂
Your friendly neighborhood beauty addict,