Skeleton point behavior identification system based on shift graph convolutional neural network and identification method thereof

Abstract

The invention discloses a skeleton point behavior recognition system based on a shift graph convolutional neural network. The skeleton point behavior recognition system comprises an image acquisitionmodule, an image processing module, an extraction module and a behavior recognition module, wherein the image acquisition module is used for acquiring a behavior image; the image processing module isused for processing the behavior image acquired by the image acquisition module for image processing; the extraction module is used for extracting skeleton points of the image processed by the image processing module; the behavior identification module is used for identifying the behavior characteristics of the skeleton points extracted by the extraction module. According to the invention, the behavior identification module is designed to identify skeleton point behaviors; the novel graph convolution of the graph convolution calculation amount is reduced; the system is different from traditional image convolution; according to shift graph convolution, a feeling range is not expanded by expanding a convolution kernel, but shift splicing is carried out on graph features through a novel shiftoperation, so that the same or even higher recognition precision is achieved under the condition that the calculation amount is remarkably reduced and the calculation speed is improved, and the situation that the calculation amount of traditional graph convolution is increased along with increase of the convolution kernel is avoided.

Description

technical field [0001] The invention relates to a skeleton point behavior recognition system based on a displacement graph convolutional neural network, relates to the general image data processing or generation G06T field, and particularly relates to the G06T7 / 20 motion analysis field. Background technique [0002] In behavior recognition tasks, subject to the constraints of data volume and algorithms, behavior recognition models based on RGB images are often disturbed by changes in viewing angles and complex backgrounds, resulting in insufficient generalization performance and poor robustness in practical applications. Behavior recognition based on skeletal point data can better solve this problem. [0003] In the skeleton point data, the human body is represented by the coordinates of several predefined key joint points in the camera coordinate system. It can be easily obtained by depth cameras and various pose estimation algorithms. [0004] However, in this traditiona...

AI Technical Summary

Problems solved by technology

However, the calculation amount of graph convolution will increase as the convolution kernel increases, resulting in a large amount of calculation for traditional graph convolution.

Images