How video AI is winning fan engagement for sports | NTT DATA

Tue, 14 July 2020

How video AI is winning fan engagement for sports broadcasters

Today, it is almost impossible to watch sport without coming across a highlights reel. Regardless of the platform, whether it’s television, social media, a streaming service, or another channel, all deliver sports highlights as news, teasers, and programs at an increasing rate. It’s hard to imagine an event or a live sports broadcast without them.

The use of highlights in sports has a long history – dating back further than younger generations would probably imagine. The “Instant replay” has been a mainstay of sports broadcasting since the late 60s, with the very first machines capable of providing immediate replays first used earlier in the decade. The technology used for CBS’ coverage of the US military's American football game – all 590kg of it – was a little clunky to say the least.

Over the course of a couple of generations, the technology, and fan expectations along with it, have changed to such an extent that they would be unrecognisable to those early pioneers of sports broadcast. Instant replays and highlight reels have become quintessential to the sports viewing experience and front and centre of fan engagement strategy. It’s not hard to see why.

In an always-on world with significant amounts of content available on the internet and insignificant consumer attention spans, highlights have become the ideal medium of choice to reach more viewers. Catering to audiences in need of instant gratification and where alternative content is just a mouse click away, the embrace of short-form content is a killer idea for broadcasters.

In some respects, though, it’s a tricky idea for them to deliver on.

Fans expect a lot from their highlights: they want highlights that are delivered more often, are more personalised to them, more convenient to watch, more analytical, more streamlined, more accurate. The bar for highlights has never been so high.

For broadcasters, reaching that bar isn’t necessarily a simple ask. The manual work required to spin up five or ten-minute recaps of live sports events is significant. There’s also human attention bias to consider. The eye will inevitably miss things that would work well in a highlights reel.

Video AI

The solution that broadcasters are increasingly turning to is artificial intelligence (AI). Commentators often note that this helps to improve the quality of edited highlights and can significantly cut down the time it takes to pull together relevant clips. But the benefits of this technology don’t stop there.

For one thing, video AI can facilitate fan consumption by simplifying the process of classifying and finding content. Whether on a news site, or a subscription streaming service, how content is delivered makes a difference to consumers and how they choose to connect with that content.

Given how each of us will have different preferences when it comes to tournaments, teams, players or plays, video AI can also personalise or curate sports content according to a viewer’s taste. Preferences can be ascertained from conditions such as viewing time or device type, and can also be established on the fly by monitoring user viewing behaviour such as pausing, replaying, and fast-forwarding content.

Video AI at The Open

Video AI is a technology that NTT DATA sees huge potential in as part of our R&D efforts, investments and partnerships, and is something that we often apply in our work with clients and partners. One of our standout use cases involves our work with The Open – the historic golf tournament – where our video AI has for several years now been integrated into a huge on-site screen which we call, The NTT DATA Wall.

NTT DATA’s award-winning Wall technology is testament to the wide-reaching capabilities of today’s video AI technologies – analysing golfers’ reactions to their shots, and body posture as well as real-time fan reactions. Out of more than 100 cameras filming at The Open, just 5% of the footage actually makes it the viewer’s TV screen. The rest sits in offline archives or is deleted without delivering any value. Our video AI technology can take these unused camera feeds, and will clip and tag to make them searchable and easily compilable.

The technology provides a whole new narrative layer to the visitor experience at The Open, and much like Hawk-Eye in tennis or VAR in football, adds an extra touch of drama for the fans.

Of course, video AI doesn’t have to have a physical presence at sports events to engage fans. Indeed, this year, The NTT DATA Wall can’t make an appearance because of the cancellation of The Open 2020. Instead, NTT DATA will be applying video AI to produce a new program called The Open for the Ages, which airs this week.

The Open for the Ages will splice archive footage stretching from 2015 right back to the 70s, pitting golf’s greatest players against each other in a kind of virtual tournament. Artificial intelligence technology will lie at the heart of the program’s production: it will support human editors in producing additional ‘shoulder content’, automatically compiling highlight clips from the virtual event and additional engaging clip packages for golf fans.

Events like this demonstrate the incredible power of artificial intelligence for the media industry during these uncertain times, particularly for sports broadcasters. As my colleague and media practice lead Ruggero Di Benedetto, explains: “digital engagement is usually quite secure across the industry, tending to be governed by a very uniform and consistent calendar. COVID has disrupted all of that, and across the sector, content vaults have been unlocked to build programs that would have otherwise been occupied by current competitions.”

Artificial intelligence, exemplified in the work that NTT DATA is doing this year, is the very tool that can give these old or previously unused assets new life.

For more information on video AI and our work at The Open, stay tuned for our ebook, releasing later this month.

How can we help you

Get in touch