#Facial beauty prediction via deep cascaded forest

Social media, networking, dating apps, and other resources, such as entertainment software, might have a use for an automated system that can analyse a photo of a person’s face and determine how beautiful that face might be to other people. Research published in the International Journal of High Performance Systems Architecture suggests that a deep cascaded forest could be the answer to developing a prediction system of beauty.

The researchers based in China and Italy have used multi-grained scanning to obtain the features of the portrait and then applied multiple random forests to enhance the person’s features ahead of classification. Tests with a data set of some 10000 previously categorised portraits showed the new algorithm developed from their approach could accurately assign a degree of beauty, automatically for the people without any eye to behold them.

“The method used in this paper is superior to other methods in feature extraction and prediction accuracy relatively,” the team writes. They add that their optimised approach will ultimately offer a stable and accurate facial beauty recognition tool. Of course, as is often remarked beauty is in the eye of the beholder and no automated approach to determining whether someone is handsome, pretty, or other is going to be considered 100% accurate when assessed by real people all of the time.

However, for a dating app or website, a rough and ready way to categorise people and so give them a more equitable opportunity to match with a potential new partner could be more successful given that real people really do judge books by their covers, however shallow that may seem.