Accurately tracking the ball in 3D space is crucial for sports analysis. Existing technologies, like goal-line technology in soccer, rely on expensive setups with multiple cameras. Our research explores using computer vision and machine learning to estimate the ball's 3D position in cost-effective, single-camera videos.

We focus on two promising approaches:

• Direct 3D Prediction: Neural networks can be trained to directly estimate the ball's 3D location from a single image, considering its size and surrounding scene. While effective, this approach can be imprecise due to inherent limitations of only considering single images.
• Physics-Guided Tracking: We can also track the ball's 3D movement across a video sequence, ensuring predictions align with the laws of physics. This method offers greater accuracy.

The Machine Learning and Computer Vision Lab investigates both techniques, with a particular interest in leveraging physical knowledge for more precise 3D ball location estimation in sports analysis.

For more information, please contact Daniel Kienzle.

References:

• Daniel Kienzle, Katja Ludwig, Julian Lorenz and Rainer Lienhart. Towards learning monocular 3D object localization from 2D labels using the physical laws of motion. International Conference on 3D Vision 2024, March 18-21, 2024, Davos, Switzerland.  DOI: 10.48550/arXiv.2310.17462