Skip to main content

Google can now recognize objects in videos using machine learning

Google can now recognize objects in videos using machine learning

/

A ‘whoa’ moment for artificial intelligence

Share this story

If you buy something from a Verge link, Vox Media may earn a commission. See our ethics statement.

Photo: Google

Fei-Fei Li, chief scientist of artificial intelligence and machine learning at Google Cloud, came on stage at Google’s Next Cloud conference today to talk about the current and next-generation applications of AI that Google’s working on. These technologies will make a difference in self-driving cars and healthcare, sure, but also Snapchat’s filters and Google Photos’ search capabilities. But the big highlight came when she announced a new way to allow software to parse video.

This new “Video Intelligence API” was demoed onstage, and it offered the kind of “whoa” moment you expect from a Google keynote. By playing a short commercial, the API was able to identify the dachshund in the video, when it appeared in the video, and then understand that the whole thing was a commercial. In another demo, we saw a simple search for “beach” and was able to find videos which had scenes from beaches in them, complete with timestamps. That’s similar to how Google Photos lets you search for “sunset” and pull up your best late-day snapshots.

Before now, computers couldn’t really understand the content of a video directly without manual tagging. “We are beginning to shine light on the dark matter of the digital universe,” Li said. At least in Google’s demo, it was genuinely impressive. And Google is making the API available to developers, just as it has with its other machine learning APIs.

The demo came near the end of a long keynote about Google’s attempt to convince everybody that it’s a serious player in the cloud services game. To signal its investment in this business, Google brought out the big guns: SVP of Google Cloud Diane Greene spoke, of course, alongside CEO Sundar Pichai and Alphabet’s executive chairman Eric Schmidt.

“We’re beginning to shine light on the dark matter of the digital universe.”

The keynote today wasn’t exactly a barn burner. It’s apparently impossible for anybody to talk about cloud services and machine learning without resorting to vague platitudes. Even if this isn’t an event meant for consumers, it was remarkably light on specifics about Google’s services. Verizon, HSBC, eBay, Home Depot, Disney, Colgate-Palmolive, and SAP also spoke about their partnerships with Google — alternately in glittering generalities and arcane enterprise software acronyms.

The big question is whether Google’s splashy attempt to make the case that it can play on equal terms with Amazon and Microsoft will move the needle at all. As The Wall Street Journal notes, Google is significantly behind those two competitors, even though its technology is competitive. Basically, the companies Google needs to woo are all already engaged in longstanding partnerships with another cloud provider. As Dave Bartoletti, principal analyst at Forrester Research, told the Journal: “They’re not invited to the party enough.”

But if Google can find practical applications for whiz-bang features like the Video API demo it showed today, it might find itself allowed into the next soirée.