During the last two decades, rapid development in the network technology has been observed, particularly hardware, and the development of software technology has accelerated, resulting in the launch of a variety of novel products with a wide range of applications. Traditional sports training systems, on the other hand, have a single function and a complex operation that cannot be fully implemented in colleges and universities, causing China's sports training to stagnate for a long time. The goal of physical education and training is to teach a specific action to attain its maximum potential in a variety of ways. As a result, we should use the system to collect scientifically sound and trustworthy data to aid relevant staff in completing their training tasks. Therefore, in the context of big data, network information technology has become the main way to improve the physical education system. By applying cloud computing technology, machine vision technology, and 64-bit machine technology to the physical education training system, extract the video data of the physical education system, design the system video teaching process, and complete the construction of three-dimensional human model, so as to analyze the training situation of the trainers. In this paper, 30 basketball majors in a university are selected as the professional group and 30 computer majors as the control group. The average reaction time, scores, and expert scores of the two groups are analyzed. The results show that the test of the professional group is significantly higher than that of the amateur group. At the same time, the feedback results of students using physical education and training system and normal physical education teaching and training are compared and analyzed. One week later, the students trained by the physical education system have improved their thinking ability, movement accuracy, and judgment ability, indicating that the application of the physical education training system to the actual effect is ideal.