LAS VEGAS — Could machine learning technology help advertisers control whether or not to buy an ad on controversial videos? Software advances suggest that may happen soon.

So far, video has remained a black box – what little data there is about its content and context tends to be added manually and is partial.

But now the machine learning titans have brought their cloud software to pitch to TV companies at NAB Show.

We have already heard how Microsoft’s Cognitive Services has a new video API, used by Ooyala, that can bring facial analysis, voice recognition, spoken word recognition, written word identification, on-screen word understanding and in-motion object detection to bear on digital videos.

And now David Mowrey, head of product in IBM Watson’s media division, no less, tells Beet.TV his machine learning tool is touting the same capabilities.

“It’s about finding all the unstructured data that’s in video, pulling that out in to something that can be used,” Mowrey says. “Once you have that, then you feed that in to a personalisation engine or your ad optimisation products, and really we can increase the value of that video.”

Mowrey says IBM’s Watson magic, when worked on video, can read and deliver:

  • text recognition

  • closed captioning

  • audio sentiment analysis

  • scene markers

  • keywords of content

  • event understanding

  • face recognition

  • character recognition for logos and text in videos

“We can leverage that both on the buy and the sell side – find audience for it and get it in front of the right audience at the right time,” he adds.

Media and entertainment is a relatively new client area for IBM to explore.

Looks like video is heading in to a new era of automatic recognition and tagging, one which could help both publishers and ad companies alike.

This video is part of Beet.TV’s coverage of the 2017 NAB Show in Las Vegas.   The series is sponsored by Ooyala.  For more coverage of NAB, please visit this page.