Like many husbands, Henry Kang often found himself pressed into service as his wife’s fashion adviser. “What can I wear with this?” his wife, Shawna, would ask each morning. Though his Ph.D. training in the School of Computer Science left him perhaps better prepared to provide coding advice, he nevertheless managed to help her.
Then Kang had an “a-ha moment” while watching “Clueless,” the classic 1995 movie in which the lead character, Cher (played by Alicia Silverstone,) used a computer program to help her choose her outfits each day. Kang realized that he could make something much better using the iPhone and modern technology.
In 2013, he founded Peekabuy Inc. with the help of some fellow SCS alums and developed a StyleIt, an iPhone app that helps users put together outfits. The company launched the app in the spring of 2014 and in December the iTunes App Store featured it as one of the best new lifestyle apps for iPhone.
After downloading the app, users simply take a photo of a piece of apparel that they plan to wear; StyleIt then suggests 10 outfits that complement that piece. A pair of shoes, for instance, will be matched to jackets, blouses, jeans and skirts.
“For the users,” he said, “it’s really magical.”
The user can complete the desired outfit by finding items in her wardrobe that closely match the suggested items.
“Of course, if she really likes the item we’ve suggested, it’s very easy for her to buy it,” said Kang, who serves as Peekabuy’s CEO. With StyleIt, people can purchase those items from one of 450 stores, including Forever 21, J Crew, Tory Burch, H&M and Urban Outfitters. All of the items can be purchased and shipped with a single click, even if they are from multiple outlets. Peekabuy receives a commission on all sales. It also makes use of the new Apple Pay technology.
The technology underlying StyleIt is image-based object recognition, the subject of Kang’s Ph.D. research in the Robotics Institute. He was co-advised by Takeo Kanade and Martial Hebert and received his Ph.D. in 2012. Two Ph.D. alums of the Computer Science Department, Hua Zhong (CS ’07) and Xi Liu (CS ’11), also helped found Peekabuy.
The Peekabuy team was able to eliminate a computational bottleneck for object recognition, making it extremely fast and efficient. To build the app’s fashion sense, they used machine learning programs to analyze tens of thousands of fashion blogs. The app continues to learn, adjusting to each user’s preferences based on which of the ten suggested outfits the user selects in each instance.
Though the team had strong technological skills, developing a consumer product such as StyleIt “is insanely harder than it would appear,” Kang said. To create the right user experience, he explained, he and his colleagues had to go out and talk to their target customers.
“That, at the beginning, was very new to us,” he said, recalling how he and his colleagues would go to coffee shops to demonstrate the app and get feedback on how to make it better.
But they also learned that interaction also could be especially rewarding.
“When you show her the app and you see her eyes light up – that’s an awesome feeling,” he said.