BeachTracker The goal of this work is the implementation of a tracking
method that allows semi-automatic positional determination
of the four players from pre-structured beach volleyball
videos. This enables the extraction of statistical measures
for video analysis and the classification of game situations
such as serves, attacks or defense. The tracking is based on
different cues, such as a color histogram approach and a
background subtraction algorithm, implemented in a particle
filter framework. The project furthermore deals with the
automatic field calibration and the ball tracking. The ball
tracking is used to determine the exact ball contact time
points. For a good technical and tactical analysis by the
trainers, it is especially the positions of the players on
the field at those time points that matter the most. The
software developed goes by the name of BeachTracker, and
is a complement to the BeachScouter and BeachViewer
software of our partners at the Technical University of
Munich. | Project manager: Prof. Dr. Björn Eskofier
Project participants: Dipl.-Ing. Gabriel Gomez
Keywords: Computer vision; particle filters; tracking
Duration: 1.6.2012 - 31.3.2013
Sponsored by: Bundesinstitut für Sportwisschenschaft (BISp)
Mitwirkende Institutionen: Technische Universität München
Contact: Gomez, Gabriel E-Mail: gabriel.gomez@cs.fau.de
| Publications |
---|
Gomez, Gabriel ; Linarth, Andre Guilherme ; Link, Daniel ; Eskofier, Björn: Semi-automatic tracking of beach volleyball players. In: Byshko, A. ; Dahmen, T. ; Gratkowski, M. ; Gruber, M ; Quintana, J. ; Saupe, M. ; Woll, A. (Ed.) : KOPS Institutional Repository University of Konstanz (Sportinformatik 2012: 9. Symposium der Sektion Sportinformatik der Deutschen Vereinigung für Sportwissenschaft Konstanz 12.-14.09.2012). Konstanz : n/a, 2012, pp 22-27. |
Institution: Chair of Computer Science 5 (Pattern Recognition)
|
|