JINCHEON — At a demonstration of SKT-FAN (Fencing AI Nexus) held at the National Training Center in Jincheon, North Chungcheong Province, on the 26th, a monitor played footage of a men's sabre bout from the Asian Fencing Championships held in New Delhi in June. Red lines marking joint movement appeared over the athletes' arms, legs and torsos, tracking their rapid motion in real time.
Each time an athlete on screen launched an attack, the system generated a timeline marking the moment of each point at six seconds, 14 seconds and 30 seconds. Below it, an artificial intelligence tactical assessment report appeared in just three seconds, covering the situation overview, key technical and tactical actions, right-of-way and the basis for the referee's call, and coaching comments. For a conceded point, the comments read that the athlete had been pushed back by the opponent's pre-emptive forward speed and reacted late, and that a response timed to the attack was needed.
Won Woo-young, coach of the men's sabre national team, reached much the same conclusion without seeing the AI analysis. "The footwork slowed down on the attack, and the athlete watched the opponent too much and went in cautiously," the coach said. "The steps should have been short and quick, with a bolder attack."
SK Telecom has set out to convert the experience and tactical intuition of coaching staff and performance analysts — long regarded as tacit knowledge in sports — into explicit knowledge in the form of data and reports. SK Telecom (017670.KS) unveiled the fencing AI performance analysis system SKT-FAN on the same day at a send-off ceremony for the national squad heading to the 2026 Aichi-Nagoya Asian Games.
SKT-FAN began as an idea from Choi Jong-hyuk, a manager on SK Telecom's sports marketing team who had been handling in-house YouTube marketing. Last year, after hearing from a national fencing team performance analyst that there was far more footage to review than any person had time to watch, the manager took in-house AI training and began developing the system directly. Fencing poses a particular challenge because foil, epee and sabre each have different rules and scoring methods, and interpreting what happens in a bout requires long experience and tactical intuition. Choi communicated directly with national team coaching staff and performance analysts over an extended period, building their accumulated analysis criteria and know-how into the algorithm.
SKT-FAN consists of score recognition and automatic slicing, motion recognition and analysis, individual athlete reports, and AI performance analysis. AI video recognition technology detects changes on the scoreboard to automatically identify the moments points are scored or conceded, then organizes those sequences into highlights and a timeline. Analyzing a single match previously took an hour; it can now be done in 10 minutes.
The system also applies sensorless motion capture. It tracks an athlete's joint points and movement using ordinary match footage alone, without attaching costly wearable sensors. The extracted motion is converted into data and accumulated in a database for each athlete. Combined with footage of past matches, the data feeds into analysis of an athlete's movement, competitive characteristics and strengths and weaknesses. SKT-FAN then pairs this with discipline-specific algorithms prepared in advance by coaching staff and performance analysts, along with a large language model, to produce the final report.
With AI handling repetitive video analysis and organization, coaches can concentrate on higher-value work such as reading matches and setting strategy. The AI recognizes scores, sequences and motion to build timelines and highlights and carry out a first round of analysis, after which specialists review and refine the results and apply them to training and strategy. "What is innovative is that the AI breaks down and explains things we hadn't thought of," Won said. "It also unpacks points that were hard to explain to the athletes, so it has become easier to talk things through while watching the footage."
SK Telecom, which chairs the Korean Fencing Federation, plans to use SKT-FAN for actual performance analysis at this Asian Games together with the federation, and to reflect the experience in functional improvements in 2027. The goal is to keep raising the system's capabilities and analytical accuracy through the 2028 Los Angeles Olympics to help improve the national team's performance.
According to SK Telecom's annual report, its sponsorship of the Korean Fencing Federation rose from 1.9 billion won in 2022 to 2.8385 billion won in 2023 and 3.055 billion won in 2024, totaling 7.7935 billion won over the three years. This year, the company is going beyond its existing support for infrastructure and competitions to apply AI technology itself to strengthening the national team's competitiveness.
