Skip to main content

New computer vision tech shows what you'd look like with any hairstyle

Dreambit: how would you look like with blonde hair/curly hair/in 1930/in india/....
When it comes to haircuts, there’s no easy way of knowing just how you’d look with a particular style. (Maybe that’s why my look hasn’t changed in roughly two decades.) Fortunately, technology is here to the rescue. It’s called Dreambit, and it can show you exactly what you’d look like with any hairstyle, or even in any historical time period.

Developed by computer vision researchers at the University of Washington, Dreambit builds off the team’s previous work in the fields of facial processing, three-dimensional reconstruction, and age progression.

As complicated as the underlying technology is, the system is designed to work as simply and intuitively as possible. Just upload your favorite selfie, type a keyword into a search box, and voila. You can type a specific hairstyle or type of hair (such as “curly”) or even search for places or time periods. Dreambit analyzes the image and scans its database of over 100,000 source files based on your search terms, then seamlessly blends images together to create a surprisingly realistic rendering of your new ‘do.

While this may sound like fun and games, there is also a much more serious application of the technology. Working alongside age progression software, it could potentially help in missing persons cases.

“It’s hard to recognize someone by just looking at a face, because we as humans are so biased towards hairstyles and hair colors,” said Ira Kemelmacher-Shlizerman, the lead researcher on the project, in a press release. “With missing children, people often dye their hair or change the style so age-progressing just their face isn’t enough. This is a first step in trying to imagine how a missing person’s appearance might change over time.”

Dreambit will show off the technology next week at SIGGRAPH. Interested users can currently sign up for beta access to the platform, which should be coming soon.

Editors' Recommendations