Home
People
Research
Publications
Events
Links
 
 

Trajectory-Based Analysis of Team Sports

Home | Tracking | Segmentation | Recognition | Evaluation | Application | Results

State of the art player motion tracking and analysis system

Our analysis system is build modularly and is divided into five modules:

  1. Calibration module
  2. Tracking module
  3. Annotation module
  4. Presentation module
  5. Analysis module (under development)

 

System facts

  • It takes approximately 20 hours of work (calibration, tracking and statistical analysis) to obtain the motion data for all 10 active players on the court for an entire basketball game.

  • The analysis of a handball game can be done in about 30 to 40 hours of work since it requires more players to be processed, there are more contacts between players and players enter and leave the court all the time.

  • Other sports such as squash or tennis which involve fewer players can be processed even faster in approximately 2 to 3 hours.

Want to learn more about our system or obtain the demos? Please contact us.

 

Calibration module

The calibration module is used calibrate each video to the same court coordinate system. During the calibration process, the starting and the ending points of the game in video and the type of the game are manually specified, and then the small number of key points is placed along the court boundaries, following the computer’s instructions. This way, the computer can determine the correspondence between the image and court coordinates.

 

 

 

  
Check the publications to find more about technical details of this module.
 
 
 
 

Tracking module

The tracking module is used to obtain reasonably accurate data about player motion around the court. The maximum RMS error of the obtained data varies from 0.3 m under the camera up to 0.5 m at the court boundaries. Players are initialized manually by mouse clicks, and then the automatic processing of video is started.

 

  
Check the publications to find more about technical details of this module.
 
 
 

 

Annotation module

Annotation module it is used to enter manual expert annotations about the events observed on the court. These annotations can be then used to produce the high-level game statistics about number and positions of different game elements such as passes, shoots, screens, etc. Since they are recorded in the same framework as player motion data, they are perfectly aligned.

 

  
Check the publications to find more about technical details of this module.
 
 
 

 

Presentation module

Presentation module is used for to view, export and print the results, obtained with help of the the tracking and annotation modules. It can be used to produce appealing visual presentations of obtained data.

 

 

  
Check the publications to find more about technical details of this module.
 
 

 

previous page | next page

 

Research sponsored by



 
 
Selected publications:  
   
(1) M. Perše, M. Kristan, J. Perš, S. Kovačič."Automatic Evaluation of Organized Basketball Activity". In: Michael Grabner, Helmut Grabner (eds.), Computer Vision Winter Workshop 2007, St. Lambrecht, Austria, pp.11-18. February 2007.

 
(2) M. Kristan, J. Perš, M Perše, S. Kovačič."Towards fast and efficient methods for tracking players in sports". In: Proceedings of the ECCV Workshop on Computer Vision Based Analysis in Sport Environments, pp.14-25. May 2006.

 
(3) M. Perse, M. Kristan, J. Pers, S. Kovacic."A Template-Based Multi-Player Action Recognition of the Basketball Game". In: Janez Pers, Derek R. Magee (eds.), Proceedings of the ECCV Workshop on Computer Vision Based Analysis in Sport Environments, Graz, Austria, pp.71-82. May 2006.  
(4) M. Perse, M. Kristan, J. Pers, S. Kovacic."A Template-Based Multi-Player Action Recognition of the Basketball Game". In: Janez Pers, Derek R. Magee (eds.), Proceedings of the ECCV Workshop on Computer Vision Based Analysis in Sport Environments, Graz, Austria, pp.71-82. May 2006.  
(5) M. Kristan, J. Perš, M. Perše, M. Bon, S. Kovačič."Multiple interacting targets tracking with application to team sports". In: 4th International Symposium on Image and Signal Processing and Analysis ISPA, (Awarded best student paper award), pp.322-327. September 2005.  
(6) M. Perše, J. Perš, M. Kristan, G. Vučkovič, S. Kovačič."Physics-Based Modelling of Human Motion using Kalman Filter and Collision Avoidance Algorithm". (4th International Symposium on Image and Signal Processing and Analysis ISPA), pp.328-333. September 2005.  

Click here for more publications.