A hybrid framework to predict ski jumping forces by combining data-driven pose estimation and model-based force...

Area of Science:

• Biomechanics
• Sports Science
• Computer Vision

Background:

• Accurate measurement of forces in ski jumping is crucial for performance analysis and injury prevention.
• Traditional methods often require markers or sensors, limiting applicability in real-world competition settings.

Purpose of the Study:

• To develop and validate a hybrid framework for predicting ski jumping forces from motion video.
• To combine data-driven pose estimation with model-based force calculation for non-invasive force prediction.

Main Methods:

• A skeletal model with five joints and four rigid segments was developed.
• Deep neural networks were used for markerless pose estimation from video data.
• Joint forces were calculated using dynamic equilibrium equations based on estimated joint coordinates.

Main Results:

• The proposed deep neural network directly estimates joint coordinates from video without markers.
• Laboratory experiments confirmed the validity and usefulness of the hybrid framework.
• The method achieved a relative error of less than 7% in maximum jumping force prediction.

Conclusions:

• The developed hybrid framework offers a practical, non-invasive solution for predicting ski jumping forces.
• The markerless approach enhances applicability in real competition environments.
• This method facilitates advanced biomechanical analysis in sports.