The informatization of education is the key area of the current education reform. Physical exercise is not only a compulsory course in schools but also very important to people's health. With the outbreak of the COVID-19 epidemic, online courses have begun to appear frequently. With the gradual increase of various online learning platforms, digital platforms emerge as the times require, which lays the foundation for the provision of physical education online. The digital platform can make online physical education teaching more complete and efficient and gradually form an independent and efficient system. However, the development of digital-physical education platforms has not made a big breakthrough. In response to the above problems, this paper will conduct data analysis and research on the role and development of digital platforms in physical education based on the Internet of Things (IoT) technology. The research results showed that the Random Forest (RF) algorithm can achieve 99% accuracy in gesture recognition when the number of features is 4. This provided new methods and aids for physical education on digital platforms.