Referees have a tough job – this French company believes AI can help

Baseballs are often thrown at 100 miles per hour (160km/hr if you prefer). The destination is an imaginary box, one with invisible boundaries that exist only in the rule book and the minds of players, referees and fans.

And while everyone squints to follow this flashing white wonder, the referee has the hardest part. After all, their job is to determine whether the ball, barely visible to the naked eye, entered the invisible box or not.

Most of the time, referees get it right. Sometimes, they get it wrong. They are human, after all. While purists accept the imperfections of refereeing as part of what makes live sports appealing, many are embracing technology as a way to enhance the quality of the sport. Defeat or victory are both better when fair.

One company bringing technology to sports grounds near you is ST37. The French company is using AI to analyse video captured as the action occurs and helping referees make better decisions in sports such as volleyball, fencing and, now, baseball. And it’s showing immense promise, having won the Best French Sportech award from Business France at CES 2024 (where we interviewed its CEO, below, edited for clarity).

Related reading: How cloud computing transformed football – even at park level

Could you please start by introducing yourself and the company?

My name is Carlos Pineda and I’m the CEO of ST37, a video analysis software company.

We use artificial intelligence to analyse video captured during sports events and determine what has happened. For example, in baseball, we can analyse the strike zone and advise a referee, enabling them to make a more informed decision.

In the example (pictured above), we are demonstrating how the strike zone calculations work. It’s very complex to achieve because we need to understand the body of the batter, the ball moves extremely fast and we need to determine everything in real-time for the referee.

In this second example (above), we need to understand many parameters; what arrived first, the ball in the glove or the foot to the base? Did the catcher have his foot on base? In this case, AI determines that the runner arrived 80 milliseconds later than the ball, information that gets passed on to the referee.

Do you work with other sports as well?

Yes. We work with volleyball, in which we can determine whether the ball touches the line or not, as well as whether the net is touched or not. We also work in fencing, in which we can advise a referee in real-time, which is particularly complicated because it all happens very fast. Our company was the first one to use AI to do so.

How long has ST37 been in business?

We started five years ago, based in the south-west of France. This year we are investing €14 million to expand internationally, notably baseball in North America.

Who are your typical customers?

Mostly sports organisations, on a licence basis. We can analyse sports matches and make the data available in an online application, which can be used for reviews, scouting and surfacing relevant information. It lives in the cloud and leverages AI to determine key points for customers.