Hive Builds Tailored AI Models via 700,000-Person Workforce

Hive, a startup founded by Kevin Guo and Dmitriy Karpman, trains domain-specific artificial intelligence models via its 100 employees and 700,000 workers who classify images and transcribe audio. The company uses the Hive Work smartphone app and website to recruit the people who label the data, and recently introduced three products: Hive Data, Hive Predict, and Hive Enterprise. Shortly after the product launch, Peter Thiel’s Founders Fund and other venture capital firms invested $30 million in the startup.

VentureBeat reports that Guo stated they founded Hive “because we felt that while there’s a lot of excitement around AI and deep learning, we didn’t see many practical applications being built.”

“There’s a lot of hype, but didn’t seem obvious what problems they’re really going to solve,” he continued. “Most of these things were demos that were somewhat working, but weren’t really enterprise-grade.” The 700,000 users, who come from 30 countries and work for “small rewards,” help “process roughly ten million tags a day with 99 percent accuracy.” Hive Data then provides tailored data-labeling services to its clients.

“Getting training data to build these models is actually really, really important,” said Guo. “It’s almost ironic in a sense that the only way to automate is by enlisting an enormous amount of human labor. You can have the best framework there is, but without good training data, you’re not gonna be able to have a good output.”

Hive Work also feeds Hive Predict, the company’s “custom-designed computer vision models for enterprises that help automate business processes, and Hive Enterprise, which targets domains like auto, retail, security, and media with customized deep learning models built from scratch with proprietary data.”

One model, for example, is Logo Model API, which detects logos and “also the products or ads on which they’re displayed and the duration they’re visible,” with 99 percent recall and 98 percent precision. According to Guo, Hive is adding 100 logos a week, “with the goal of reaching 10,000 by Q4 2018.” With a backend based on Google’s TensorFlow, Hive develops its AI systems “via an API or the cloud, or engineers an on-premises solution in partnership with integration partners.”

Hive has thus far “created machine learning models that recognize activity, predict age and gender, classify cars, determine the distance between a camera sensor and a subject of interest, and even detect things like explosions, gunshots, fights, and commercials in television feeds.” Guo will not release the names of its clients, “but said that each is making tens of millions of API requests a month.”

No Comments Yet

You can be the first to comment!

Sorry, comments for this entry are closed at this time.