IBM To Introduce Watson-Powered Video Analysis This Year
By Jef Cozza / Data Storage Today. Updated April 24, 2017.
A new cloud service by IBM aims to help companies extract metadata from raw video footage, the company said today. The new tool uses the artificial intelligence (AI) capabilities from IBM’s Watson platform to analyze video for information such as keywords, main concepts, visual imagery, tone, and emotional context.
"The service highlights IBM's continued focus on combining artificial intelligence with the IBM Cloud to help media and entertainment companies make sense of unstructured data and make more informed decisions about the content they create, acquire and deliver to viewers," the company said. Although the new service does not yet have a name, the company said it is planning to launch sometime later this year.
Automatic Highlight Reel
The new platform will use several of Watson's existing APIs, such as tone analyzer, personality insights, natural language understanding, and visual recognition. In addition, IBM said the platform will include new technology from IBM Research that can analyze data produced by Watson to split videos into discrete scenes based on cues in the content.
As an example, IBM said a sports network could use the service to identify content that is related to basketball and contains positive or exciting scenes based on the language, images, and sentiment included in the video. The platform would allow the network to automate the process of identifying each segment and editing them together, allowing for much faster turnaround.
Advertisers could also use the platform to analyze several years of content from a television show, enabling them to generate and cut together a collection of clips focused on particular themes such as footage of the characters driving in a car or sitting around the kitchen table.
Bringing AI to Video Analysis
Media and entertainment companies could use the platform to help manage their content libraries. For example, a company might want to prioritize content that targets viewers who want uplifting stories about world adventures.
IBM said the platform is particularly relevant given the current media environment, in which video content is proliferating at an exponential rate and being delivered across a number of different devices. This growing volume of content is making it increasingly difficult for companies to analyze video in a timely fashion using human editors.
The new platform builds on an earlier project IBM has been pursuing to integrate its Watson AI and other cognitive technologies into its cloud video solutions, according to the company. Last year, the company used experimental versions of some Watson APIs to edit together a movie trailer. In that experiment, Watson analyzed trailers for other movies to learn how to make an effective one. It then created an entirely new trailer for a movie that had not yet been released.
IBM said it has also been working with the U.S. Open to use Watson to convert audio commentary into text more accurately by teaching the AI tennis vocabulary.