Automatic Ball Detection in Soccer, Golf and Tennis Games

Status: open
Supervisor: Robert Sablatnig

Problem Statement
Detecting and tracking moving objects is a challenging task, since the objects’ optical signature is less discriminative due to occurring motion artifacts, occlusions in the wild and dynamic backgrounds.

The goal of this practical training project is to explore and examine methods that can be used to automatically detect balls in various sports (soccer, tennis, golf) and to estimate the accurate 3D trajectory in a region of 20 meters length and 7 meters height. One or two cameras at a fixed position are used to capture the object 0.5m or 1m after the performed pass, to avoid occlusions by the player’s body parts. The required precision for ball trajectory estimations should lie within the range +-3cm and for the computation of velocity within +-1km/h. Mono-/Stereo-cameras and different illumination setups should be evaluated for indoor and outdoor acquisitions.

Literature Review – getting to know the algorithms
Data Preparation
Written Report/Thesis and final presentation