IBM serving up ace AI at Wimbledon 2019

Share this post:

Wimbledon is the tennis tournament every player dreams of winning and one of the highlights in the sporting calendar.  As the world’s leading tennis tournament their brand focus is “in pursuit of greatness” demonstrating their continual desire to strive to be the best while also acknowledging that is a continual journey.

For the last 30 years, IBM has supported Wimbledon in that aspiration with the latest AI and cloud technologies to deliver award-winning digital experiences to fans and media around the world. To stay ahead, Wimbledon needs to deliver even better insights and content that highlights the tension and excitement about The Championships in order to entertain a vast and growing global audience of fans and broadcasters.  With this focus Wimbledon now consider themselves to be a data driven media business as well as a world class sporting event.

Wimbledon as an early adopter of Artificial Intelligence (AI) have been turning new data sources into competitive advantage helping then deliver with capabilities from new fan experiences to IT operations. For several years now IBM has used machine learning on the data to unlock the “keys to the match”. The patterns and styles of play that will help players optimize their performance against specific opponents. Presented to fans via the IBM SlamTracker on Wimbledon.com these help fans understand how the key tactics within the game are developing and how the match was won.

The fuel for AI is data. Since 1990 IBM has collected nearly 63 million data points from the tournament. How do you gather the right data with 100% accuracy from each court? The answer is to place skilled tennis players courtside to assess and record details of every point. It takes a real expert to reliably spot more subjective points like an unforced error. This partnership between man and machine is a common pattern and crucial element to effectively using AI. We go to such lengths to ensure the data we provide is trusted by press and media from around the world. They base their stories on it so it must be right. Detailed statistics captured for every point in the match, recorded in real time using specialist software and hardware. We target 100% accuracy sub-second.

Media-savvy fans want – and now expect – to see the most exciting moments as quickly as possible. Thanks to groundbreaking AI systems developed by IBM, we’re now able to capture the best bits from multiple games and matches at any one time. With over 18 courts and 4 matches per court per day, hundreds of hours of footage are produced.  Previously, finding and editing video highlights was done manually. Now hundreds of hours of video content are analysed by innovative AI technology allowing highlights to be selected within 2 minutes of a match finishing.

IBM’s Watson AI technology continuously tracks the action, ranking every point, measures the emotion of the crowd via noise levels, reads the gestures of the players as well as analyzing game play and statistics about the matches. It’s this combination of factors that help identify the real highlights that audiences are interested in.  All this means the fastest and most comprehensive highlights are curated and available more quickly to fans and to the media.

As a learning system, one of the benefits of using AI is the ability to continue to easily enhance the solutions with new capabilities. For Wimbledon this year Watson has been taught to better recognise acoustics and understand inadvertent bias increasing the quality of the output.

As a business where integrity and equality of service to all players is a core principle providing equality of highlights to all is essential. Not all highlights during a tennis match are equal. For example, a highly passionate crowd favourite could produce more indicators of an exciting point than a more reserved, yet equally skilled, opponent.

Using Watson OpenScale for AI, the system can now analyse many different data points introducing additional fairness and bias reduction in the selection of notable video for highlights. For example, a Gentleman’s Singles match on Centre versus a Girls’ Doubles match on Court 7 will have very different attributes – for example: camerawork, audience size, player emotion, etc. OpenScale will automatically account for these differences through machine learning and provide results which are fair between the 2 different types of matches.

In 2019 Watson Acoustics will be “listening” to the strike of the ball, adding richness to the data sources to help create even tighter highlights packages.  Sound analysis from the system can also now recognise when the ball has been struck allowing tighter and more accurate editing of highlight clips saving vital time and maximising every second of footage rights.

For Wimbledon, the ability to scale their AI capability to serve their audiences is critical to their strategy; last year, the highlights contributed to over 14 million video views.  We see that scalability as critical to business too; AI can help business leaders be more productive and produce better results – at scale.  Wimbledon’s infrastructure and systems scale up to 55,000% compared to a typical day outside of the tournament fortnight, so the flexibility of the IBM Cloud is perfectly placed to ace the requirement come July each year. You can read more about it here.

The story of AI at The Championships in 2019 is a continuation of a trusted 30 year partnership.  In their pursuit of greatness, Wimbledon have shown continued commitment to innovation, and IBM are proud to provide the expertise and thought-leadership they need to turn data into insights for all who love the tournament.

Engaging tennis fans by providing a tailored and trusted experience is similar to the challenge faced by many of IBM’s clients as they seek to find ways to engage their customers and provide them with the information and service they demand. Royal Bank of Scotland (including NatWest and Ulster Bank) give their customers access to a self-service virtual assistant called Cora that helps with many common enquiries, reducing wait times and providing faster customer service. Cora which was initially trained with over 1,000 responses to more than 200 customer queries, constantly learns. When Cora can’t help the customer is seamlessly passed to a live advisor who has all the relevant information right in front of them. Cora is powered by IBM Watson Assistant.

Find out more about how IBM’s Watson AI technology can help your business