Custom AI for Sports

World Cups. League tournaments. Grand Slams… It’s safe to say that the buzz around sports is never ending. However, despite media being the industry’s biggest revenue generator, Mckinsey’s analysis shows that in the 2016–2017 regular season, NFL ratings declined 9%. The decline in ratings was caused by a drop in the number of games watched and the minutes watched per game. Brian Hughes, of MAGNA Global USA believes that, “increased interest in short-term things, like stats and quick highlights… has funneled some young viewers away from TV.” Content producers and distributors are waging a battle against decreasing viewer attention spans and only those who strategically employ innovative technologies can hope to be front-runners in the industry.

PwC projects that the revenue generated from sports in 2019 for North America will be $73.5 billion given that in the United States alone, 70% of the adult population claim to follow sports regularly.

By leveraging proprietary player and object tracking algorithms AI offers content producers the exciting opportunity to support high-speed sports broadcasting enabling them to create multiple versions of the same content suited to the consumer’s liking. During most sporting events, multiple live feeds capture the same match at different angles generating approximately 10 hours of content for every 1 hour of an actual match. And the content producers are left to manually analyze and extract relevant content from this bulk data. With the help of technology and Artificial Intelligence (AI), content creators and distributors can curate sports content with utmost ease. As an example, AthenasOwl (AO) has leveraged the power of AI and has developed several use cases that can weed out these inherent inefficiencies. Some of these are

  1. Quick highlights: Use of AI can minimize the turnaround time with which broadcasters can get back to the audience with a highlight video
  2. Custom highlights: Use of AI can also create custom highlights for different user segments depending on fanbase, interest areas, region etc., allowing broadcasters to improve customer targeting, ratings and revenue.
  3. Content monetization: Each complete coverage video of a match can be analysed and repurposed into multiple versions of content, thereby monetizing the large amount of data generated. This means, for example, that from 1 hour of raw footage, multiple clips can be created with one focusing only on the high points of the game, another on players’ close-up shots and yet another on crowd reactions.

The technology that AO is built upon, breaks up given raw content into micro moments, then tags these moments and stitches them together to create custom content. Let’s take an example to understand this better.

Germany’s Mario Götze scores the match-winning goal in the 2014 FIFA World Cup finals between Germany and Argentina

Meta tagging this moment that marked Germany’s win in the 2014 FIFA World Cup involves collating information on five important aspects.

  1. Camera angle detection: The moment is tagged with information regarding whether the frame is shot from an aerial view, as a closeup or at field angle.
  2. Custom location detection: Information on the primary subject of the frame is captured. In this case the focus is on the key players while in other cases, it could be the crowd, the ball or the field.
  3. Logo detection: The logos on the players’ jerseys or on the field are detected depending on the camera’s angle.
  4. Custom keyword detection: After analyzing the background commentary, specific keywords that are relevant to the moment are extracted.
  5. Face recognition: Most importantly, AI can also collect information on players whenever the player’s face is captured at suitable angles.

While techniques of tagging and analysis of sports media have been employed by various AI entities, the combined use of these techniques to create custom highlights is a novel venture. Working on the principles of hyper-tagging and ‘moment intelligence’, AO can accurately decipher the raw data offering multiple opportunities for clients.

Sports media is a unique playing field for AI technology given the abundance of information created during each event. Streamlining the analysis of this data and applying it in a financially viable manner is potentially game changing. AO is a pioneer in this regard. By integrating the detection of multiple facets of information ranging from camera angle detection to face recognition, AO can dissect media content into numerous points of value. The final products intelligently assemble these moments to allow users to access multiple highlights of a single game, or to select from multiple perspectives of a live match making the viewing experience more enjoyable.

Comments

Leave a Reply

Your email address will not be published. Required fields are marked *

Previous Article

Theory of Moments

Read more

Next Article

Content Analytics: Decoding Audience preferences

Read more