In 2016, Vinayak Shrivastav and his team of technology specialists started building a smart system that can take any live feed and cut it up into short-form clips. Within those clips, the AI software is able to identify specific object and player-related movements. Additionally, the AI can identify commentator language and map all outputted information to various searchable keywords. This allows scouts and sports teams alike to scour thousands of hours of game film in a minimal amount of time, producing videos of impactful moments.
Shrivastav, the CEO and co-founder of Magnifi, said the technology is able to freely sift through both audio and visual inputs from videos, which provides a high level of customization.
“Let's take soccer there, so while you're watching a soccer game, the goals scored, yellow cards, red cards, no penalty, fouls, free kicks, substitutions — so there are multiple moments that are predefined while you're watching a sporting event,” Shrivastav said. “We've trained it to identify these defined sporting moments. It can also be a fight between two players, it can just be an argument, it can just be exchanged a few words between two players or the referee, anything of that sort that interestingly catches an eye, we have trained the machine to identify those specific moments.”
The AI is able to identify the start and end points of particular interactions and impactful moments by searching for body movements and listening to the commentator's words and tone.
“We are able to identify what the players are wearing,” Shrivastav said. “So seeing the jersey numbers, we're able to have data feedback into the system where we understand what player is having a moment there. If they have scored a goal, we’re identifying has the ball touched the net, or has it crossed the line? So if you're able to identify these, we understand that there's an inflection point to the moment, and then we are able to identify a key moment that is attached to that particular event.”
Shrivastav noted that a majority of people nowadays are used to watching content in short form, citing the mass use of applications like TikTok and Instagram, where videos typically run between 15 seconds to one minute. Due to the sheer amount of content existing on the social media marketplace today, he believes many people end up missing live events and instead catch up after the fact.
“A lot has changed — specifically after the COVID came in,” Shrivastav said. “A majority of the people have now started consuming content online, sporting viewership has never been the highest in the last year or so. And what we have realized is that not everyone is going to be able to consume the entire game in real-time, but the short-form videos have a much larger shelf life and also a larger impact for the audience.”
One important way the Magnifi AI software is being utilized is to help scouts find professional players for their teams. Regardless of the sport, many scouts are forced to rummage through high amounts of footage from previous games and practices. With keyword searching and the ability to customize the inputs for a desired video output, scouts are able to find players who frequently excel at specific actions.
Shrivastav said being able to focus on players’ deliberate and precise actions in a game lets coaches and recruiters more aptly construct their teams.
“Imagine you're basically the scout, and you need to draft a couple of key players, and you have a ton of footage that you need to go through — it's practically impossible for you to go through everything and you're not going to be able to scout by going to each of the locations,” Shrivastav said. “So if you have all the video files in one single place, and you're looking for specific things like you're looking for players that have had the best outcome in, say, two-point shots, or three-point shots, you're able to kind of filter those out. And to see those particular moments and then make a decision.”
The Magnifi AI system also has the ability to learn from the keywords you assign to specific moments. Once a player is identified, a play type is analyzed, a movement is chosen or a sound is parsed out, the AI will add all similar moments to a library. This function makes it simple for players, teams and broadcasters to find exactly what they are looking for right when they need it.
Currently, the biggest clients of Magnifi are broadcasters, sports teams and federations, individual players and team scouts. In some cases, college athletes are beginning to use the program to create their own highlight reels.
“Over the last 18 months, we have just started seeing more traction coming in from sports teams, sports federations,” Shrivastav said. “That's when we realized that there's a larger use case there because they're just not limited to using this as an engagement. They want to monetize it, they want to create short-form content to insert advertisements, or have it on social.”
Magnifi is finding use from everyday consumers making online content to professional broadcasters who use the software to actively search and cut up short clips during live games.
Shrivastav said that his motivation for developing the Magnifi AI came from a love of sports at a young age. He grew up playing soccer and always wished someone was recording him so he could go back and view those plays at any time.
Shrivastav noted that Magnifi has the potential to solve a content generation problem while reducing cost and time dedicated to production.
“The immediate problem solving is taking a live footage and how can you edit the live footage without actually requiring a video editor,” Shrivastav said. “Because the biggest problem today is you need to make someone sit behind the computer screen to edit the content, which is not efficient, it's expensive.”
A major impact point for Magnifi came during and after the COVID-19 pandemic when many people were stuck at home.
“People wanted to consume more content because literally everyone was sitting at home,” Shrivastav said. “So everyone was going to consume content, so they needed volume. Now volumes cannot be created by individual human beings, I think that's when we actually started seeing growth and traction kick in actually, where people genuinely understood what the problem was. And it can only be solved by a machine by automating a bunch of processes. The creativity cannot be taken away from a human and the machine can never be creative enough. But at least it can solve the volume problem.”
Magnifi is looking toward the horizon for what their AI is capable of. In the future, they see uses in other areas of the entertainment industry, vehicle dash cam analytics and text-to-video content generation.
http://accesswdun.com/article/2023/2/1170489/artificial-intelligence-helping-sports-teams-catalog-game-footage