Basketball free throw hit probability prediction system based on deep learning object detection and pose estimation

With the rapid progress of artificial intelligence, deep learning technology has been extensively adopted in sports. As a key scoring point in basketball games, accurately predicting and analyzing the trajectory of arm movements is crucial for enhancing the hit rate of free throws. Therefore, to improve the prediction probability of free throws in basketball games, a spatial transformation module is used to optimize the extraction accuracy of object detection in densely populated sports venues. In addition, the study also utilizes a three-dimensional pose estimation algorithm to predict the vector field of joint key points and joint connections. The accuracy in predicting the angle of the shooting arm and the angle of the take-off leg reached 92.06% and 91.12%, respectively. In summary, the designed system has effectively improved the hit probability of basketball free throws.