Frequent visitors know that in every salon in America and around the world, there are expert persons who know anybody’s perfect hairstyle. They can choose a perfect haircut, style, or hair dye that complements the customer’s face.
But what if in the near future, we have a software for our computers or smartphones that can predict the best hairstyle or hair color for our beloved face? What if before you could even leave the comfort of your home for your weekend salon makeover, an application could already show you the perfect cut?
That isn’t actually the prime direction of a new algorithm software from the University of Washington. It’s called Dreambit, and the main goal of its developers is to help predict the future look of a person.
It came from the previous research conducted at the said university and elsewhere for facial recognition. In Dreambit, the combined algorithm delivers “a unique way” to create the blended images, says the university in a press release.
Anyone can now upload their photo at Dreambit.xyz. After uploading the file, a user can change the appearance of the face in the photo by searching terms such as curly hair, 1930s for the ‘throwback’ look, or India for a look with eastern vibe.
The software changes the appearance of the person by mining internet photo collections that are similar to the search term and seamlessly map the person’s face onto the results.
The new software, the university said, can help show a missing child or person’s appearance if years have passed. It can also help locate people evading the law if they’re in disguise.
University of Washington assistant professor Ira Kemelmacher-Shlizerman with her team previously developed the automated age progression software that only focused on a person’s face. With the Dreambit, the team has added varied hairstyle options and other elements.
Kemelmacher-Shlizerman explains that we humans are biased towards hairstyles and hair colors. Hence, it’s hard to recognize someone by just looking at a face. With missing children and people, she said that people often change their hairstyle or dye their hair, so the algorithm for age-progressing isn’t enough.
In addition to finding missing people, or suggestion for somebody’s next hairstyle, this software could also help the Hollywood casting market.
It can “visualize” how the actor would look like in a role, and Kemelmacher-Shlizerman says it’s a way to try on different styles or personas without actually changing the actor’s physical appearance. As an example, the developers have used Cate Blanchett mixed with Bob Dylan in the movie ‘I’m not there.’