The 2018 World Cup in Russia is underway, and this year, the on-field refs will have some help to offset human error with the goal of preventing bad calls.
In March, the FIFA Council approved the use of Video Assistant Referees. This team of 13 refs will be based in control rooms in the soccer stadiums and will monitor 33 video feeds, some of which operate in super-slow motion. The VAR will help make decisions on difficult calls in “match-changing” situations. These are incidents that have the potential to change the outcome of the game, depending on the ref’s call.
Tim Ouillette, an assistant teaching professor at Northeastern, said that part of the reason FIFA is implementing VAR technology in the World Cup this year is to bolster fans’ faith in the referee system. This is important, given the mishaps that occurred at the 2014 World Cup, which included missed penalty calls and offside confusion that cost Colombia a goal.
“FIFA is trying to legitimize themselves, reclaim the ethos of credibility, and make the right call,” said Ouillette, who is a filmmaker and videography expert.
This isn’t the first time video technology has sought to change the game for sporting events. Raymond Fu, an engineering professor at Northeastern, noted that even more advanced video systems are already used in other sports.
For example, machine learning algorithms and video processing are used in tennis to measure the speed of a serve, said Fu, who holds joint appointments in the College of Engineering and the College of Computer and Information Science.
The algorithm is able to recognize what’s happening in the game and learn about players based on the speed with which they hit the ball and where it is placed on the court. Referees can use the video as a reference system, reviewing plays from multiple angles. For players, algorithms offer predictions and strategies based on the system’s analysis of the tennis ball’s speed and accuracy.
Could this type of artificial intelligence be used by refs to analyze soccer plays? Not yet. According to Fu, it would take much more computational power to apply the same machine learning technology to games in real-time.
“I don’t imagine even the most advanced technology could achieve performance results that recognize events and human activities in real-time,” Fu said.
Soccer is an inherently fast-paced sport with many moving pieces, Fu said, which is why making accurate calls can be such a challenge.
“You’re talking about a group of humans playing together, not just one person—and they may have very different juxtaposed speeds,” said Fu. “Even following the ball is tricky because it could be anywhere on the field.”