AI-powered Athlete Tracking Makes Olympic Debut

In elite competition, fractions of a second can mean the difference between a gold or a silver medal. The Tokyo 2020 Olympics is using a new tracking technology that allows fans to watch the action and know how fast the athletes are running. With the help of Intel® Xeon® scalable processors, certain data points are made available to broadcasters to help them provide more insight into athletes’ performances.

Image credit: Intel Newsroom/EXOS video

Game on. One year after the pandemic delayed athletes’ dreams, the Olympic Games Tokyo 2020 start in July, and these will prove to be the most technologically advanced Olympic Games ever. From the airport to the event venue to the TV screen, artificial intelligence will be prevalent in many areas, improving the experience for athletes and viewers alike.

One of these innovative technologies is 3D Athlete Tracking from Intel, an AI-based system used to analyze certain data points. 3D Athlete Tracking technology, or 3DAT by Intel, leverages the processing power of Intel® Xeon® Scalable processors, cameras, and computer vision to capture certain near real-time performance data from athletes during their events.

The 3DAT technology follows 22 locations on the athlete’s body. By using cloud computing from Alibaba Group and Intel hardware, artificial intelligence platforms and computer vision will analyze skeletal data, body angles, speed, and other performance metrics to determine stride length, cadence, top speed, and speed at given moments during a race.

Some data, such as speed at given moments, will be visible to fans as part of post-event analysis using on-screen overlays. Racers and speed at given moments will be identified through color indicators, similar to a heat map, moving from yellow to red, getting darker as they go faster. Data will then be made available after the event to broadcasters who can opt to show the additional insights gathered by 3DAT. The clips will be available to Olympic broadcasters within about 60 seconds for replays.

The solution requires a lot of processing power. For example, the 3DAT technology will track nine sprinters simultaneously, each running for 10 to 12 seconds. The video data is sent to the cloud at 60 frames per second, where it is analyzed using Intel® Xeon® Scalable processors with integrated Intel® Deep Learning Boost AI acceleration capabilities. AI can fill in the gaps when runners are crowded or overlapping.

Intel partnered with Ashton Eaton, two-time Olympic champion of the decathlon, to help develop the technology.

Competitive Boost

Not related to the Olympics, the 3DAT technology is used in other ways and is being tested by EXOS, an elite health and training company with operations across the US. EXOS is using 3DAT as a coaching tool to help football players and other athletes who want to go pro gain a competitive edge. Because the process relies on cameras and AI processors, athletes don’t need to wear special suits or sensors.

Using multiple cameras (or smartphones or other devices), coaches record athletes during drills and training sessions. 3DAT provides athletes with insights about their biomechanics, speed, stride length, velocity, and acceleration when running or sprinting. Working with a trainer, the athlete can make modifications or adjust workouts to improve performance. Trainers can view reports and charts that detail the athlete’s body mechanics and performance and drill down to identify potential problems.

graphic depiction in foreground of runner metrics with runner in background

Image credit: Intel Newsroom

The 3DAT technology unlocks data that was previously not measurable by the naked eye. Coaches and runners traditionally had to rely on feel to tweak performance. The precise measurements from 3DAT eliminate the guesswork and allow players and coaches to make data-based precision adjustments. And for the first time, we’ll be able to see just how fast world-class athletes can run.