After decades of controversy and disputes surrounding judging, figure skating is turning to artificial intelligence (AI) in an effort to achieve greater transparency in how performances are evaluated.
Over the past two years, the International Skating Union (ISU) has been testing high-resolution camera systems at competitions. These systems utilize AI to track skaters' movements and analyze technical elements such as jump rotation, height, distance covered, and spin positions in real-time.
The computer vision technology, which employs six cameras positioned around the rink, aims to provide more objective data to support technical panels. These panels currently make split-second judgments on whether jumps have the correct rotation or from which edge a competitor took off – decisions that can ultimately determine medal outcomes.
ISU Director General Colin Smith stated that the initial goal is to use the data to assist judges in awarding technical scores, and then potentially integrate it into the actual scoring system.
The ISU plans to develop the system for individual competitions first, followed by pairs and ice dance – the discipline most affected by judging controversies.
"We are always looking for where we can incorporate technology and where we can develop the sport to ensure fair competition on the ice," Smith told Reuters. "If it's effective, judges will be able to focus on the artistry, on the human element, and computer vision will focus more on the technical, practical aspects," he added.
Omega, the official timekeeper for the Milan-Cortina Olympic Games, is using computer vision to gain a better understanding of performance, and the company stated it could "see the data it collects as an additional tool for judges in the future."
One such type of data comes from a new system that can detect the angle of each skater's blades, providing judges with additional information on how well certain routines and jumps are executed.
The ISU also plans to use data analysis to evaluate the performance of judges. It has analyzed over 750,000 element scores and nearly 270,000 component scores from 78 international events to identify specific issues.
"Judging requires a lot of skill. We want to be able to identify who the best judges are and then develop them further, and then look at those judges who might be in the lower tier and help them develop their skills. What we are looking for is consistency in judging across different events," Smith further commented.