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

Sitting posture recognition method based on projection reconstruction and multiple-input-multiple-output neural network

A technology of neural network and recognition method, which is applied in the direction of neural learning method, biological neural network model, character and pattern recognition, etc. It can solve the problems of difficult modeling, too sensitive camera angle changes, poor anti-interference, etc., and improve the recognition accuracy Effect

Active Publication Date: 2020-06-23
NANJING UNIV OF TECH
View PDF4 Cites 4 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] The present invention proposes a sitting posture recognition method based on projection reconstruction and multiple-input multiple-output neural network (MIMO-CNN). And defects such as being too sensitive to changes in camera angles, the depth of sitting images of the human body are obtained through the depth camera, and preprocessing and 3D information reconstruction are performed on them to obtain multi-view depth information. The identification of the sitting posture state and the sitting posture state in the left and right directions, the user can feed back misjudgment samples during use, and regularly relearn and optimize the network parameters, thereby improving the recognition accuracy of the network

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
  • Sitting posture recognition method based on projection reconstruction and multiple-input-multiple-output neural network
  • Sitting posture recognition method based on projection reconstruction and multiple-input-multiple-output neural network
  • Sitting posture recognition method based on projection reconstruction and multiple-input-multiple-output neural network

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0095] The present invention will be further described below in conjunction with the accompanying drawings. The following examples are only used to illustrate the technical solution of the present invention more clearly, but not to limit the protection scope of the present invention.

[0096] A sitting posture recognition method combining projection information and multiple-input multiple-output neural network (MIMO-CNN), the flow chart is as follows figure 1 shown. Include the following steps:

[0097] Step 1, the specific implementation process of image acquisition is:

[0098] A1, use the depth camera to obtain the depth image, such as figure 2 ;

[0099] A2, use the random decision forest classification algorithm of the official SDK of the depth camera to finally divide the human body into 32 parts, and finally get the human body foreground contour we need, such as image 3 ;

[0100] Step 2, the process of image preprocessing is:

[0101] B1. Perform histogram avera...

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 sitting posture recognition method based on projection reconstruction and a MIMO-CNN. The sitting posture recognition method comprises the following steps: acquiring a humanbody upper body depth image and a human body foreground contour map; performing pretreatment; projecting the depth information of the sitting posture contour, and reconstructing to obtain a three-view depth map; designing an MIMO-CNN network for sitting posture recognition and learning model parameters; sitting posture recognition; model self-learning. The method has the advantages that the preprocessed depth image is combined with the human body contour map, and interference of surrounding backgrounds on sitting posture recognition is eliminated. A three-view depth map is obtained by using aprojection reconstruction method, so that sitting posture information is richer. The invention discloses a designed MIMO-CNN structure. The method is especially suitable for projecting and reconstructing feature information and integrating an attention mechanism at the same time, hot spot areas of different sitting postures can be better concerned, so that the recognition precision is improved, meanwhile, model self-learning is adopted, the requirements for real-time performance and accuracy are well balanced, and the method has high anti-interference capacity for visual angle changes and complex environment backgrounds.

Description

technical field [0001] The invention relates to a sitting posture recognition method based on projection information and a multiple-input multiple-output neural network (MIMO-CNN), belonging to the technical field of human posture recognition. Background technique [0002] With the rapid development of science and technology, sitting posture has already become one of the most common daily states of modern people, and it is also closely related to the human body. Most of our writing and office work and in front of the computer are carried out in a sitting posture, especially the sitting posture of teenagers and children is not standardized during the learning process, but few people will notice the impact of sitting posture on health. There are a lot of bad habits when using the computer, like bowing your head, hunchback, sitting on your back, sitting obliquely, etc. Therefore, it is of high practical value to automatically recognize the sitting posture of a person through a...

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): G06K9/00G06K9/62G06K9/60G06N3/04G06N3/08
CPCG06N3/08G06V40/20G06V10/20G06N3/045G06F18/214G06F18/241
Inventor 沈捷黄安义王莉曹磊
Owner NANJING UNIV OF TECH
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