Human motion recognition is an important issue and a well-reported issue for future human development. Because human movements have different meanings, or different movement postures and precision problems in different sports, using artificial identification and judgment, there will be individual differences or artificial judgment errors. For future development, such as the development of human-computer interaction VR system, as well as medical rehabilitation, surveillance care, sports training, and other developments must rely on human motion recognition, and there is a further development. However, identification through the information system is a major problem, so it extends and develops different identification methods. For example, image recognition, accelerometer, gyroscope identification or thermal image recognition, etc., also collect more detailed sports information of the user through the wearable device, as a basis for judging, giving the user the basis and training for training. In different types of sports, Tai-Chi Chun is a slow exercise training. In the past, research pointed out that Tai-Chi Chun is beneficial to improve the cardiopulmonary function and the recovery of the motor function of the elderly. However, training Tai-Chi Chun requires self-training through coach training. Finally, it is hoped that the wearer-type device and mobile device will assist the user in training Tai-Chi Chun self-test and training. Based on the collection of experienced Tai Chi coaching data, we design a recognition system that is fed back to the user.