Badminton Technical Training Analysis System Based on Intelligent Motion Tracking Decomposition Model

Abstract

Although scientific research in the field of sports has achieved fruitful results, the relevant research only presents a fragmented technical normality. In view of the fact that the achievements of professional sports technology can not be applied and compared, it is necessary to use a mature intelligent analysis system to collect and analyze the information of the whole process of project training and competition. Considering the dependence of badminton on technical and tactical movements, there is a need to judge the body movements in various training and establish a motion model to provide technical support for ensuring the accuracy of movements. Based on the enhanced training data set, this paper innovatively uses a lightweight convolutional connection deep neural network to establish a three-dimensional structure data model according to the badminton limb movement law. Intensive training is carried out with the samples processed by single-cycle sampling. A high-quality action trajectory output effect is obtained to correct the wrong trajectory in the movement. Through experimental analysis, the model parameters are updated on the enhanced training set, which verifies that the research results of this paper have achieved good results in the calculation accuracy and response-ability. Compared with the traditional algorithm, it has the advantages of a lightweight structure and obvious prediction effect. It is especially suitable for embedded platforms such as intelligent robots, which provides technical support for sports and makes some contributions.