Movement Analysis and Action Optimization of Physical Education Teaching Practice Based on Multisensing Perception

Abstract

Correct and effective physical education teaching can not only improve students’ physical quality but also exercise students’ willpower, which is an important content to promote students’ all-round development. However, according to the current teaching situation in our country, in the actual teaching process, there is a situation of incongruity between teaching and sports development, which leads to the decline of the quality of physical education teaching in our country and affects the development of students’ comprehensive quality. Based on these problems, starting from the relationship between teaching and sports, this paper analyzes the coordinated development between physical education teaching and training in colleges and universities and builds a physical education teaching quality monitoring system. The research results of this paper show that (1) when using traditional recognition of various motion patterns, it can recognize various behavior patterns, and the average recognition accuracy is 90.1%. The accuracy is 94.3%. Compared with the traditional recognition mode, the average recognition accuracy is increased by 4.2%, and the recognition result is better. Compared with the recognition results of the first set of experiments, for the more difficult to distinguish upstairs and downstairs, the recognition accuracy is increased by 9% and 7%, respectively, and the recognition accuracy of backward is increased by 6%. (2) Before receiving the teaching, the test results of each index of the members of the routine group and the training group were basically the same, and there was no major difference. After the $T$ -test was performed between the conventional group and the training group, the results showed that the $P$ values of the evaluation results of the two groups were both above 0.05. The experimental results showed that the initial conditions of the two groups could be regarded as the same before receiving the teaching. Combining the evaluation results of the two groups before the training, we can conclude that under the condition that the initial conditions are basically the same, and the training conditions and environment are basically the same, the trainees who have received the mode training method have obtained better physical fitness indicators. The improvement and the effect are greatly optimized compared with the mode training. (3) Among the 8-spoke images captured by the experiment, the multisensor motion analysis model proposed in this paper has the highest action recognition accuracy. When the first picture is taken, the recognition accuracy is 98%. The recognition accuracy rate is also increasing, and when the eighth image is taken, the action recognition accuracy rate reaches 99%. Among the three different models, the multisensor motion analysis model proposed in the article has the shortest page response time. When the number of tests is 10, the average page response time is 0.4 seconds. When the number of tests increases to 70, the average page response time reaches 1.0 seconds, and the success rate of the multisensor motion analysis model has remained at 100%. The average response time will increase with the increase of the number of tests, and the experimental results also show that the detection performance of the multisensor motion analysis model is the highest.