Kinematics of ski alpin worldcup slalom racers by using AI-technologies and recalibrated cameras

(Kinematik von Weltcup-Slalomfahrern im alpinen Skirennsport unter Verwendung von KI-Technologien und rekalibrierten Kameras)

INTRODUCTION: Knowledge of the kinematics of alpine ski racers is of great interest. Even at the elite level, the work of Reid (2009, 2010) in slalom and Spörri (2012) in giant slalom showed differences. But in their experimental design with several rotatable and tiltable cameras, only two complete turns could be analyzed. The aim of the method presented is to analyze the entire race course and all competitors. METHODS: The runs of the 30 best starters at the two WCs were filmed in both rounds by four video cameras from different positions. By using an AI software provided by SIMI ®, selected landmarks (joint points) on the filmed runner could be detected automatically. In addition, the positions of the turning gates were measured (noVR ® and FIS service) and, through the knowledge of a sufficient (= 7) number of control points, were used for the subsequent calibration of the cameras. The necessary number (= 2) of further control points to determine the tilt, pan and zoom of the cameras is also given by the gates in the pictures. This allowed the utilization of the (re-) calibrated cameras to determine 3D coordinates from the land points determined by the AI. RESULTS: To check the accuracy, the thigh lengths were determined after the first round of calculation. There were relative standard deviations of about 10%. Further results will be presented at the congress. DISCUSSION/CONCLUSION: First results show that the presented method makes it possible to determine the joint points of the racers during an entire run. If a sufficient quality is reached, a number of interesting analysis options are enabled, e.g. turn cycle structures, trajectory lengths, center of mass speed over the turn cycle, center of mass attack angle, center of mass inclination, hip/knee angulation, ... (Reid, 2010). In the next step, the position of ski (tip and heel) and poles should be detected by extending the AI Algorithm. In this case, calculating of ski orientation (ski edge angle, ...) would be possible.
© Copyright 2023 9th International Congress on Science and Skiing, March 18 - 22, 2023, Saalbach-Hinterglemm, Austria. Veröffentlicht von University of Salzburg. Alle Rechte vorbehalten.

Schlagworte: alpiner Skisport Biomechanik Tracking Video Technologie
Notationen: Ausdauersportarten Naturwissenschaften und Technik
Tagging: Slalom künstliche Intelligenz Kinematik
Veröffentlicht in: 9th International Congress on Science and Skiing, March 18 - 22, 2023, Saalbach-Hinterglemm, Austria
Herausgeber: T. Stöggl, H.-P. Wiesinger, J. Dirnberger
Veröffentlicht: Salzburg University of Salzburg 2023
Seiten: 107
Dokumentenarten: Kongressband, Tagungsbericht
Artikel
Sprache: Englisch
Level: hoch