A Study on Synergistic Improvement of Physical Fitness and Skills of Basketball Players Based on Big Data

Abstract

This paper proposes a strategy for synergistic improvement of physical fitness and skills of basketball players with the content of big data analysis of decision-making of physical fitness training programs and estimation of the human posture of basketball players. The data related to physical training and testing are processed using neural network methods, and a priori algorithms with continuous attributes discretize the data to realize data fragmentation and prevent data distortion caused by intrinsic correlation so as to build up a technical and methodological platform for physical training programs. At the same time, the basketball action posture estimation algorithm based on multi-scale spatiotemporal correlation features is proposed, and the human body temporal sequence feature capture module based on Transformer is constructed to improve the accuracy of basketball action posture estimation to meet the needs of skill training. The men’s basketball team of Shandong Agricultural University in China was used as a research object to carry out basketball physical fitness and skill training practices. The experimental group’s basketball players’ performance in barbell bench press and weighted squat was 11.06kg and 10.05kg more than that of the control group. Their performance in 3/4-court sprint running, and return running. Restriction-area footwork was 0.52s, 2.16s, and 1.97s faster than that of the control group, with significant differences in basketball skills, such as 14-meter round-trip straight-line dribbling around the pole and 30-second in situ spotting of two-handed chest pass (the first time in the experimental group). Basketball skills showed significant differences (P<0.05).

Keywords

• neural network
• a priori algorithm
• multi-scale spatiotemporal correlation features
• transformer
• basketball training
• 62-07