Abstrait

Basketball training via behaviour analysis in basketball match video

Li Lun

In recent years, basketball match video analysis and video mining has received much attention, and it also can help player to predict the behavior of ball carriers. This paper proposed a novel method for play training by analysis the much video which has cluttered background, the low resolution of the head images and fast motion of the sportsman. This paper proposes the adoption of a covariance descriptor to min the video features of the head images, and finishes the head pose classification by the trained model. According to the video analysis and trained model, we can predict the behavior of each player, such as pass route or shot. Based on this model and video analysis, we can help player training more accurate and efficient. Finally, the tests on the basketball match verify theeffectiveness of our model.