Basketball Shot Prediction Using 2nd Order Polynomial Regression

Monitoring basketball is the most important and difficult functionalities in sport’s analysis because of how fast the game moves. The purpose of this study is to describe a computer vision-based method for detecting ball features, tracking the trajectory of ball, and estimating the likelihood that the ball will strike the basket. On the video frame depicting the sporting event, HSV-based color segmentation is utilized to locate and identify the basketball. The mask that contains the features of area and centroid is used to detect the shape of the ball. It is necessary to trace the path that the ball travels in order to gather information regarding the exact position of the ball in each respective frame. The second-order polynomial regression is utilized to produce predictions regarding the trajectory that is followed by the ball as well as the likelihood that it will contact the base. The actual path and the predicted path that was followed are almost identical. The accuracy of the proposed method is 97.98%.