What is it about?
This paper presents SynthNet, a tool for tracking 3D tennis ball paths using a single camera. It detects hits and bounces, then calculates the ball’s path using synthetic data, avoiding issues found in traditional methods. Tests show that our physics-informed SynthNet is more accurate and stable than existing approaches.
Featured Image
Photo by John Fornander on Unsplash
Why is it important?
SynthNet makes tracking 3D movements, like tennis ball trajectories, more accessible and affordable. Traditional methods often require expensive multi-camera setups and complex calculations, limiting their use to high-budget environments. By enabling accurate 3D tracking with just a single camera and synthetic data, SynthNet could make advanced performance analysis available to more athletes, coaches, and researchers, improving training insights and expanding the possibilities for sports science and data-driven coaching.
Perspectives
Many neural network approaches are merely a block-box. SynthNet uses knowledge from the real world via physics-informed calculations.
Stella Grasshof
IT University of Copenhagen
Read the Original
This page is a summary of: SynthNet: Leveraging Synthetic Data for 3D Trajectory Estimation from Monocular Video, October 2024, ACM (Association for Computing Machinery),
DOI: 10.1145/3689061.3689073.
You can read the full text:
Resources
Contributors
The following have contributed to this page