Research/Tracking visually-similar interacting targets in sports

Since 2003 I have been working with the Machine Vision Group at the Faculty of Electrical Engineering, University of Ljubljana, where we have developed an application for tracking and analysis of human motion in sports:

Sports tracking Demo

More on sports analysis at:

http://vision.fe.uni-lj.si/research/SportA/index.html.

Video demonstration of a local-motion-based tracking:

Visualization of a color-based particle filter:

Aa0.png

Automatic generation of a virtual camera from the tracked positions of players in sports:

Image12.jpg

Video stabilization in sports:

00002.jpeg

Tracking multiple targets from a bird’s-eye-view camera:

00003.jpeg

Tracking person’s hands:

Particle filter, NCV dyn. model, 50 particles, color-based visual model, (14 failures)
Particle filter, two-stage dyn. model, 25 particles, color/motion-based visual model, (0 failures)


Relevant Publications: