Unlock instant, AI-driven research and patent intelligence for your innovation.

Human body interaction action recognition method based on skeleton features and slice recurrent neural network

A technology of cyclic neural network and action recognition, which is applied in the field of pattern recognition and computer vision, and can solve problems such as incomplete extraction of interactive information and lack of inter-frame dependency information.

Pending Publication Date: 2021-02-05
JIANGSU UNIV
View PDF0 Cites 1 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0005] In order to solve the problems of incomplete extraction of interaction information and lack of inter-frame dependency information in action recognition, the present invention proposes a skeleton-based interactive spatio-temporal modeling method on the basis of single-person skeleton graph convolution. information to increase the accuracy of action recognition by capturing this additional interactive information

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Human body interaction action recognition method based on skeleton features and slice recurrent neural network
  • Human body interaction action recognition method based on skeleton features and slice recurrent neural network
  • Human body interaction action recognition method based on skeleton features and slice recurrent neural network

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0052] The present invention will be further described below in conjunction with accompanying drawing.

[0053] Such as figure 1 As shown, the interactive recognition method based on skeleton features and sliced ​​cyclic neural network in the present invention mainly includes the connection of different joints within and between frames, the extraction of joint features by spectral graph convolution, and the method of sliced ​​cyclic neural network. The implementation method of the present invention will be described in detail below from these aspects.

[0054] Skeleton extraction is performed for each action sequence through OpenPose. Extract the information (x, y, z) of 15 joint points of the human skeleton in each frame of each video, where x is the abscissa of the joint point on the image, y is the vertical coordinate of the joint point on the image, and z is the joint point confidence value. The joint point connection is divided into single-person connection, interactiv...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The invention discloses a human body interaction action recognition method based on skeleton features and a slice recurrent neural network, and the method comprises the steps: obtaining a skeleton sequence through OpenPose for each action, obtaining action features through the skeletons, designing interaction actions between the skeletons, and obtaining additional interaction information to improve the accuracy of action recognition; constructing a new skeleton diagram through the connections, carrying out the approximation through a high-order fast Chebyshev polynomial of spectrogram convolution, and extracting the action features; in order to enhance extraction of time domain information, innovatively applying a slice recurrent neural network to video action recognition to capture wholeaction sequence dependence information, wherein an advanced feature map of space-time modeling can properly compensate for long-term dependence loss caused by the slice network. According to the method, the accuracy of interaction identification is improved, the applicability is good, and the speed of long-time sequence feature extraction can also be improved by using the slice recurrent neural network.

Description

technical field [0001] The invention relates to the technical fields of computer vision, pattern recognition, etc., and in particular to a human body interaction action recognition method based on skeleton features and sliced ​​cyclic neural networks. Background technique [0002] Video-based interactive behavior recognition has high practical value and broad application prospects. The goal of human action recognition is to analyze and understand the actions and interactions between people in videos. Although action recognition methods based on RGB video or optical flow have high performance, they are easily affected by changes in background, illumination, and appearance, and extracting optical flow information also requires high computational costs. Nowadays, more and more people are researching on skeleton data. The human skeleton can well represent the movement of the human body and is beneficial to analyze the movement of the human body. [0003] At present, relatively...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
IPC IPC(8): G06K9/00G06K9/62G06N3/04
CPCG06V40/20G06N3/044G06N3/045G06F18/24
Inventor 成科扬吴金霞毛启容
Owner JIANGSU UNIV