Physical exercise quality evaluation method, device and equipment and storage medium

Abstract

The embodiment of the invention provides a physical exercise quality evaluation method, device and equipment and a storage medium, and relates to the technical field of behavior recognition. The evaluation method comprises the following steps: S101, acquiring a video containing the motion process of a specific physical exercise, and segmenting the video into a plurality of video stages according to the type of the physical exercise; S102, inputting the plurality of video stages into the backbone network model to obtain a plurality of feature representations corresponding to the plurality of video stages; S103, constructing a regression network model, wherein the regression network model is trained by adopting a method of dividing a total demand based on a difficulty coefficient into training labels; and S104, inputting the plurality of feature representations into a regression network model to obtain a score of a motion process in the video. The backbone network model ensures that action features in the scoring process can be complete, and semantic information is rich enough. The regression network model greatly improves the performance of the action quality evaluation method.

Description

technical field [0001] The present invention relates to the technical field of behavior recognition, in particular, to a method, device, equipment, and storage medium for evaluating the quality of sports. Background technique [0002] Most of the scoring of sports is done manually. In some sports where there are subjective factors in scoring, such as diving and gymnastics, it is easy to cause differences in scoring standards due to personal factors of referees, resulting in unfair scoring. [0003] The action quality assessment method is to extract meaningful information from the video data, and then quantify and evaluate the quality of the action (scoring) in different ways, so as to achieve the purpose of measuring the quality of the action. [0004] The action quality evaluation methods in the prior art generally use the feature extraction network in the field of action recognition as the backbone network, such as C3D, I3D, P3D, etc., and then use the features with spati...

AI Technical Summary

Problems solved by technology

[0005] The present invention provides a method, device, equipment, and storage medium for evaluating the quality of sports, so as to improve the problem of incomplete sampling of video actions to be evaluated in existing evaluation methods in the related art

Images