Complexity Graph-Based Multilabel Classification Method of Human Action in Rope Skipping Scene

Abstract

Aiming at the problem of insufficient accuracy of multilabel classification of human action at present, a multilabel classification method of human action in the rope skipping scene is proposed. It realizes feature recognition and classification by collecting human action features in the scene of skipping rope movement and uses RNN to optimize the human action feature recognition algorithm. On the basis of feature recognition, the characteristics of human movement in the rope skipping scene are classified, the confidence map of the key point position is obtained by using the Gaussian modeling method, and the action multilabel classification is realized. Finally, experiments show that the multilabel classification method of human action in rope skipping scene has high accuracy and fully meets the research requirements.