Motion Field to Predict Play Evolution in Dynamic Sports Scenes

The paper will appear in IEEE Conference on Computer Vision and Pattern Recognition (CVPR) 2010, CA, San Francisco, USA.

We propose a method to predict locations where the game (soccer) play evolves using motion field on the ground. This method can be used for automated broadcasting system where computer-controlled PTZ cameras are recording dynamic live sport scenes.

For more details :
http://www.cc.gatech.edu/cpl/projects/playevolution/

The authors for this paper are:
Kihwan Kim (Georgia Institute of Technology), Matthias Grundmann (Georgia Institute of Technology), Dr. Ariel Shamir (The Interdisciplinary Center), Dr. Iain Matthews (Disney Research Pittsburgh), Dr. Jessica Hodgins (Disney Research Pittsburgh) and Dr. Irfan Essa (Georgia Institute of Technology)

Narration : Tucker Hermans (Georgia Institute of Technology)

Category:

Science & Technology

Tags:

• Computer Vision
• Tracking
• Pattern Recognition
• Play Evolution
• Dynamic Sports Scene
• Soccer

[kihwan23]

@roidroid : Thank you, and it is a good point, actually in the paper, we described a relationship between context of given dynamic scene and global movement of players(crowd), since the players themselves have the best view and clearest understanding of the scene, and their movement on the field reflects their interpretation and their intent (possibly)

[roidroid]

Excellent. With obvious implications for the study of crowd psychology.