Human skeleton action recognition method

Abstract

The invention discloses a human skeleton action recognition method, and relates to a method for identifying graphics. The method is a human skeleton action recognition method combining space-time attention and a graph convolution network. Diversity and complementarity of different feature information are fully mined; the weight value of each joint point of the space structure and the importance ofeach frame of the video sequence are adaptively adjusted by using an attention mechanism; the motion recognition of the human skeleton is carried out by using the graph convolution network, and the defects that in the prior art, space-time feature information cannot be better captured, and errors are likely to occur in the recognition of the difficult motion of the human body are overcome.

Description

technical field [0001] The technical solution of the present invention relates to a method for recognizing graphics, in particular to a method for recognizing human skeleton movements. Background technique [0002] In recent years, with the wide application of video capture sensors and the continuous development of human pose estimation algorithms, and its potential applications in intelligent video surveillance, patient monitoring systems, human-computer interaction and virtual reality, human action recognition has received more and more attention. more attention. Human action recognition based on machine vision is to add action type tags to videos containing human actions. The purpose is to analyze and understand the actions of individuals in the video and the interaction between multiple people. [0003] According to the type of input data, human action recognition is divided into RGB video-based methods and skeleton video-based methods. Compared with RGB images, human ...

AI Technical Summary

Problems solved by technology

[0007] The technical problem to be solved by the present invention is to provide a recognition method for human skeleton movements, which is a recognition method for human skeleton movements that combines spatio-temporal attention and graph convolutional networks, and fully taps the diversity and complementarity of different feature information , use the attention mechanism to adaptively adjust the weight value of each joint point of the spatial structure and the importance of each frame of the video sequence, use the graph convolutional network to perform the action recognition of the human skeleton, and overcome the existing technology of the action recognition method of the human skeleton , both have the defect of not being able to better capture the spatio-temporal feature information, and it is easy to make mistakes in the recognition of difficult movements of the human body.

Images