More than 1.5 billion people around the world are expected to watch the 2026 FIFA World Cup finals.
With so many fans carefully watching every pass, tackle and goal, referees are under enormous pressure to make the right decisions.
To help them, FIFA is using advanced artificial intelligence and computer vision technology to improve the accuracy and speed of officiating.
One of the key technologies being used is Sony’s Hawk-Eye system.
It supports several important tools, including video assistant referees (VAR), goal-line technology, semi-automated offside decisions and a feature that helps determine which player touched the ball last before it went out of play.
At the heart of the system is computer vision, a type of AI that allows computers to understand and analyze images and videos.
During each match, 16 special cameras are positioned around the stadium. These cameras constantly record the movement of players and the ball from different angles.
The AI system has been trained using millions of images and video clips. By studying so many examples, it learns how to recognize players, identify body parts and track the ball as it moves around the field. It can even follow players and the ball across multiple camera views at the same time.
Using many cameras is important because a single camera can be blocked by players or provide a misleading angle. Multiple viewpoints allow the system to calculate the precise positions of players and the ball in three-dimensional space. This process, called 3D reconstruction, can be completed within seconds.
The system also works incredibly fast because it has a very specific job. Unlike general AI systems that need to recognize many different objects, these models focus almost entirely on soccer-related information. They mainly need to identify players, their positions and the ball. This narrow focus allows them to process huge amounts of data quickly and efficiently.
According to FIFA, the tracking cameras collect more than 150 million pieces of tracking information during a single match. Handling such an enormous amount of data has become possible thanks to recent advances in deep neural networks and graphics processing units, or GPUs.
Deep neural networks are AI systems inspired by the way the human brain processes information. They are particularly good at recognizing patterns in large amounts of data. GPUs, which were originally developed to process images and video games, provide the computing power needed to train and run these advanced AI systems.
The same technologies are already being used in other sports. Similar systems help make line calls in tennis, determine first downs in American football and review certain decisions in basketball.
Beyond sports, these technologies have even wider applications. Self-driving cars use similar computer vision systems to identify vehicles, pedestrians and other objects around them. Related systems could also be used in security and surveillance applications where cameras need to track activities across large areas.
Despite all these technological advances, the human element remains central to soccer. AI can help determine whether a player’s toe was offside or who last touched the ball, but it cannot capture the emotions, drama and unpredictability that make the World Cup one of the world’s most beloved sporting events.
