EUREKA Eurostars E! 8661 CLASS, Hego Trac AB

Important results from the project

We have fulfilled the project’s goal to enhance the data content of ChyronHego’s player tracking technology by capturing not only the players’ 2D ground trajectories but also the positions of body parts such as arms, legs, torso and head. This has been done using computer vision in an entirely unobtrusive manner and with a minimal hardware footprint at the arena. The resulting technology has been demonstrated during a live professional sporting event.

Expected long term effects

This limb tracking technology is key to maintaining ChyronHego’s position at the forefront of sports tracking on the international stage. The step from center-of-gravity tracking to a much richer representation of sports players, enables a new generation of products based on animations and statistics. Such products include 3D animations for second screen, broadcasting or coaching purposes, with virtual camera placement anywhere on the field during highlight clips or expert analysis.

Approach and implementation

The solution developed uses cutting edge deep learning technology from the research community to perform skeleton pose estimation. It builds on the underlying technology of ChyronHego´s Tracab Image Tracking System, which tracks the position of every player to within 10cm. A GPU-powered deep neural network is then applied to analyse image patches containing cut-outs of players. This is performed from multiple cameras, which facilitates the reconstruction of the full 3D skeleton.