Skeleton behavior recognition method and device based on body part level

Abstract

The invention relates to the technical field of computer vision, in particular to a skeleton behavior recognition method and device based on a body part level. In the prior art, the accuracy of behavior identification is low, and the requirements in the actual use process cannot be met. In order to solve the above problem, the invention provides the skeleton behavior recognition method and devicebased on the body part level, and the method comprises the steps: obtaining first skeleton three-dimensional coordinates of a target object in a to-be-recognized video, wherein the first skeleton three-dimensional coordinates comprise a first difference three-dimensional coordinate, a first relative three-dimensional coordinate and a first joint point three-dimensional coordinate; based on the first skeleton three-dimensional coordinates, determining the probability that the target object is in multiple preset skeleton behavior types through a pre-trained skeleton behavior recognition model; and taking the preset skeleton behavior type of which the probability is greater than a preset threshold as a skeleton behavior type corresponding to the target object. By using the method of the invention, the behavior identification accuracy can be effectively increased.

Description

technical field [0001] The invention relates to the technical field of computer vision, in particular to a skeleton behavior recognition method and device based on a body part level. Background technique [0002] Behavior recognition is an important and challenging task in computer vision tasks, and has a wide range of applications in security monitoring, intelligent video analysis, human-computer interaction and other fields. With the development of human pose estimation technology, behavior recognition based on human skeleton has received extensive attention in recent years. [0003] Due to the natural graph structure of the human skeleton, the existing human skeleton behavior recognition methods mainly use the graph convolutional network as the main framework, but the accuracy of behavior recognition is low, which cannot meet the needs of the actual use process. [0004] Therefore, how to propose a method for improving the accuracy of behavior recognition is a technical ...

AI Technical Summary

Problems solved by technology

[0003] Due to the natural graph structure of the human skeleton, the existing human skeleton behavior recognition methods mainly use the graph convolutional network as the main framework, but the accuracy of the behavior recognition is low, which cannot meet the needs of the actual use process.

Images