Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

Lightweight graph convolution human skeleton action recognition method based on channel attention

A human skeleton and motion recognition technology, applied in the fields of computer vision and deep learning, can solve problems such as joint interaction difficulty, failure to represent joint topology and connection relationship, modeling, etc., to achieve elimination of strong dependencies and small deployment feasibility , to avoid the effect of calculation volume

Active Publication Date: 2021-07-13
TONGJI UNIV
View PDF5 Cites 3 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, such a data organization cannot represent the topological structure and connection relationship between joints, and the mutual influence between joints is difficult to model effectively.

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
  • Lightweight graph convolution human skeleton action recognition method based on channel attention
  • Lightweight graph convolution human skeleton action recognition method based on channel attention
  • Lightweight graph convolution human skeleton action recognition method based on channel attention

Examples

Experimental program
Comparison scheme
Effect test

Embodiment

[0050] Such as figure 1 As shown, the present invention discloses a lightweight graph convolution human skeleton action recognition method based on channel attention, comprising the following steps:

[0051] S1: Obtain the skeleton sequence information of the human skeleton in the video image, specifically including:

[0052] S11: Modeling and preprocessing the human skeleton in the video image to obtain initial skeleton sequence information;

[0053] S12: Obtain the first-order information, second-order information and third-order information in the initial skeleton sequence information by using the difference between adjacent frames;

[0054] S13: Under the condition that the initial dimension of the data remains unchanged, the first-order information, second-order information and third-order information are fused and added to obtain the final skeleton sequence information.

[0055] Among them, in step S11, the human skeleton in the video image is modeled by attitude estim...

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 relates to a lightweight graph convolution human skeleton action recognition method based on channel attention. The method comprises the following steps: S1, skeleton sequence information of a human skeleton in a video image is acquired; S2, joint point index information is added, and an adjacent matrix of skeleton joint points is calculated; s3, spatial features of the skeleton sequence are extracted by using an adjacent matrix and a residual GCN network introducing a channel attention mechanism, wherein the adjacent matrix is subjected to random inactivation processing during use; S4, frame index information is added, and maximum pooling processing is carried out; s5, time features of the skeleton sequence are extracted by using a first-order CNN network; s6, a final feature sequence is generated through maximum pooling, and an identification classification result is obtained. Compared with the prior art, the method has the advantages of being high in robustness, real-time performance and accuracy and the like.

Description

technical field [0001] The invention relates to the fields of computer vision and deep learning, in particular to a method for recognizing human body skeleton actions based on channel attention with lightweight graph convolution. Background technique [0002] Human action recognition is an important research direction in the field of computer vision, with broad application scenarios and market value, such as: abnormal behavior monitoring, user behavior analysis, etc. Skeleton sequence data is an abstract human motion data, which uses 3D coordinates, joint index and joint connection relationship to represent the movement of various key parts of the human body. In the early skeletal action recognition methods, most of the manually extracted features were used to process and integrate the data by means of feature mapping. After the rise of deep learning methods, the use of neural networks to model the spatio-temporal information of skeletons has gradually become mainstream, an...

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
Patent Type & Authority Applications(China)
IPC IPC(8): G06K9/00G06K9/62G06N3/04
CPCG06V40/20G06N3/045G06F18/214Y02D10/00
Inventor 刘成菊党荣浩陈启军张恒
Owner TONGJI UNIV
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic, Popular Technical Reports.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com