August 29, 2017 | Written by: Noah Syken
Categorized: AI/Watson | Events | Media and Entertainment
Share this post:
This year, nearly 75 percent of all Internet traffic will be video. It comes from social media. From corporate marketing departments. And from television broadcasters and Hollywood studios. It is rapidly becoming the primary communications medium of our time.
Analyzing mountains of video footage
The only problem is, video is not easily searchable. It is made up of sound and images, unstructured information that defies the capabilities of traditional, programmable computers. So, for this year’s US Open tennis tournament – the highest attended sporting event in the world – we’re using Watson to help analyze and curate the mountains of video footage produced during the two-week event.
At the US Open, there are as many as 18 matches going on at one time. The video streams alone could fill a small data center. And it would take an army of USTA editorial staff to watch and select all of the most meaningful moments from this vast expanse of unstructured data.
Finding highlight-worthy moments
To solve this, IBM iX and IBM Research are using Watson to identify the points and shots in a match that are “highlight worthy.” The system uses cognitive algorithms to select exciting moments from US Open match video based on video, audio, and scoring data.
For example, researchers taught Watson to understand the content of tennis video, such as detecting what a player celebrating with a fist pump looks like. It analyzes the audio to identify cheering and high levels of crowd noise. And it understands the vital statistics of the match – like game points, set points, and match points – to recognize how critical a moment is to the outcome of the match.
Once the individual highlights for the match are determined, the system uses meta-data to automatically generate the graphics and facilitate storytelling. These highlights will be shared on the US Open Official platforms like Facebook, Instagram and YouTube. And they will added to a player’s bio page on the app and website.
Watson Cognitive Highlights: beyond sports
The Cognitive Highlights functionality is part of a suite of Watson solutions called Watson for Media, that includes transcription services and photo tagging, as well. Despite the name, this suite is not designed solely for media companies. It can also be used by any corporate marketing department that pumps out whitepapers, videos, podcasts and social media every day.
Additionally, it can be used to scan security footage, or observe patterns of behavior in retail environments. It can even be used to help translate and analyze customer service calls that are “recorded for quality purposes.” Anywhere video, audio, and imagery are used, Watson for Media can help.
In addition to Watson for Media, the US Open relies on a Cognitive Concierge app that uses Watson’s Conversation API to help guests navigate dining, shopping and other service options around the venue. And the SlamTracker app provides real-time analysis of more than 41 million data points to help fans gain insight and understanding into each match.
The US Open is a large, complex event that generates massive amounts of structured and unstructured information. With Watson, the USTA can refine that data and serve it up to fans in the form of insight.
To learn more about IBM’s global sponsorships, visit ibm.com/sports