System built at Kanku Ice Arena using location tracking and AI image analysis, enabling accurate filming of specific skaters

Tokyo, Japan, March 13, 2024 — MIXI, Inc. has developed an auto-tracking camera system that uses positioning and AI image analysis technology to help improve athletes’ performances.
This system consists of an indoor positioning system (IPS) installed at Kanku Ice Arena, a national training center for figure skating, along with technology that uses real-time video to AI-predict skaters’ positions and automatically track and film them. Since February 2024, designated athletes have begun using it in training to improve their competitive performance in figure skating.

■Auto-tracking camera system installed in Kanku Ice Arena

To date, we have focused on developing technologies for video transmission and editing, including commercial applications made possible by obtaining a local 5G radio station license, as well as AI-powered automated editing of race video for publicly-managed betting sports. In recent years, we have also been developing technologies that use an IPS, which we have applied to race video of publicly-managed betting sports to provide a more immersive viewing experience. This type of location-based technology can also be applied to a wide range of sports.

A new initiative using this technology is the IPS we installed in Kanku Ice Arena which has been used to create an auto-tracking camera system for the space. The system consists of 16 locators (location antennas) installed in the arena, which assess the location of athletes wearing location tags. In addition, a PTZ (pan-tilt-zoom) camera tracks and films an athlete while they are skating, and AI analysis of the footage is performed in real time to predict their next move. Accurate automatic filming is able to be achieved by this combination of location tracking and AI image analysis. This technology is capable of detecting the fast, complex movements of athletes and can automatically track and capture footage of a specific athlete even when others are skating in the same rink.

This technology has made it easier for up-and-coming figure skaters who train at Kanku Ice Arena to analyze their own performance. The captured video displays information on the athlete’s location, path of movement, and speed, allowing the athlete to review such data.

▼Footage of training taken by the auto-tracking camera system
https://www.youtube.com/watch?v=HsEoUsvMNv8

A presentation on this technology is scheduled for Day 1 of the MIXI TECH DESIGN CONFERENCE 2024, an engineering and design conference to be held on March 18 and 19. The presentation will be streamed online and open for anyone to watch. A demonstration of this system will also be exhibited at the venue.
・Development of an Auto-Tracking Camera for Figure Skating — Positioning Data + AI Image Analysis (Japanese only)
https://techcon.mixi.co.jp/2024/d1-3.html

■Future Developments

This system was introduced as part of an initiative to improve competitive performance in figure skating. It is currently used mainly for post-practice review, but we believe that by further updating the system itself, it will also be possible to use data collected from positional information and other sensors to generate information that can lead to further improvements in the athletes’ performances. We also expect that using this system for remote coaching will help improve coaching capabilities.
We believe this system, which can detect the movements of a specific athlete among multiple athletes and respond to quick movements, can be applied not only to figure skating but also to other competitive sports and live events. By expanding it to a variety of uses in the future, we will contribute to improving athletic performance across sports and advancing the viewing experience.