SKT Puts Fencing AI Coach to the Test in Real Competition; Cuts Match Analysis Time from 1 Hour to 10 Minutes

SK Telecom has unveiled an AI coach that reduces fencing match analysis time from one hour to approximately ten minutes. The 'SKT-FAN' system, unveiled on the 26th, analyzes players' movements solely via video—without the need for body sensors—and automatically extracts scoring sequences.

The AI translates its findings into tactical language. It compiles data on approach distances and whether attack timing was early or late into text, enabling coaches and athletes to view and understand the same footage from a shared perspective.

On-site, a comparison was arranged between the AI’s assessment and a human coach’s analysis of a scoring scene by Oh Sang-uk. The AI dissected the flow of Oh’s counter-attack, launched immediately after his opponent’s failed attack, focusing on attack timing and movement patterns. National team coach Won Woo-young arrived at a similar tactical assessment, albeit expressed differently. "It’s become easier to communicate with the players because the AI explains and interprets aspects we hadn’t previously considered," Coach Won said. "It is a great help in explaining in detail the players’ subtle step changes, movements, and tendencies."

Athletes Song Sera and Oh Sang-uk cited time savings as the most significant change, noting that the system now allows them to identify missed moments as well as opponents' weaknesses.

Development of SKT-FAN began earlier this year, followed by beta testing in May; the team has been using it in earnest since this month. SK Telecom is using the Aichi-Nagoya Asian Games as its first real-world testbed to verify the system's performance and utility in actual matches and training. The company aims to refine the technology by the 2028 Los Angeles Olympics, with the goal of automating more than 90% of repetitive video analysis tasks.