Google Cloud democratizes machine learning access with AutoML Video

LAS VEGAS—Bringing together the staff, resources and expertise necessary to apply machine learning models toward analyzing video content libraries isn’t so easy for all media companies. But Google Cloud may have leveled the playing field a bit.

The company today announced AutoML Video, a new machine learning feature within its Cloud AutoML product that launched in January 2018. AutoML Video, which is still in beta, allows Google Cloud users to customize models that automatically classify video content with self-defined labels. It’s designed for media and entertainment businesses looking to simplify tasks like automatically removing commercials or creating highlight reels. It can be used in other verticals as well for video analysis geared toward tracking traffic patterns or overseeing manufacturing processes.

Google offers a pair of artificial intelligence products for analyzing video libraries.

RELATED: NAB Show 2019 spotlights tech leaders from Google, Intel

AutoML Video Intelligence has a graphical interface for training video classification models. It’s built for projects that need custom labels which aren’t covered by the pre-trained Video Intelligence API.

The Video Intelligence API has pre-trained machine learning models that automatically recognize objects, places, and actions in stored and streaming video.

At the NAB Show, Google Cloud also announced that it’s working with Viacom to apply its machine learning capabilities to Viacom’s content creation and distribution workflow. Google will help Viacom automate production of short-form clips identifying what’s relevant to a viewer, serve more contextual ads on direct-to-consumer platforms and automate content tagging, discovery and intelligence for more than 65 petabytes of content.

John Honeycutt, head of telecommunications, media and entertainment for Google Cloud, said the set of tools his company is offering in the cloud are about taking friction out of the supply chain and creating speed, transparency and efficiency.

“All of this data will flow much faster because we’ve taken all the gates out of the process,” Honeycutt said.