Body posture training method

Abstract

The invention relates to a body posture training method. According to the method, after action key frames of a coach and action key frames of a learner are collected, action video data of the coach and action video data of the learner are collected, then a DTW algorithm is used first to make an action video of the coach and an action video of the learner consistent in duration and time sequence, afirst feature plane set based on the action key frames of the coach and a second feature plane set based on the action key frames of the learner are established, and the action key frames of the coach are matched with the action key frames of the learner through comparison of an included angle between normal vectors of the first feature plane set and an included angle between normal vectors of the second feature plane set; and then an included angle between edge vectors in the second feature plane set is acquired through detailed coordinates of body joint points and is compared with an included angle between edge vectors in the first feature plane set, and the action similarity of the learner is confirmed. The method has the advantages that the training and exercise effect of the learneris improved, and actions are safe enough without hidden danger of health; and the learner can obtain enough training experience from one action, so that the training effect of the learner is improved.

Description

technical field [0001] The invention relates to the technical field of motion capture training evaluation, in particular to a human body posture training method. Background technique [0002] Human body pose recognition refers to the process of recovering the position of human body joints in an image or video based on a video or image. According to the image type, human pose recognition algorithms can be divided into algorithms based on depth maps and algorithms directly based on RGB images. Because the depth map algorithm has high requirements on image acquisition equipment, resulting in high cost, it is only suitable for high-level applications. Human body posture recognition algorithms based on RGB images have low equipment requirements, especially nowadays hardware equipment manufacturers provide a A TOF radar, the so-called time-of-flight 3D imaging sensor, continuously sends light pulses to the target, and then uses the sensor to receive the light returned from the ob...

AI Technical Summary

Problems solved by technology

At present, there is no human body posture recognition training method based on RGB images. In general, signal emitters are placed on the human body to track the posture of the human body. The cost is high. This kind of sports teaching equipment can only be used in the national training center class. Professional high-level applications are thousands of miles away from ordinary people

Images