Supercharge Your Innovation With Domain-Expert AI Agents!

Human body key point detection method and system, electronic equipment and readable storage medium

A detection method and key point technology, which is applied in the direction of instruments, biological neural network models, character and pattern recognition, etc., can solve the problems of insufficient natural information and limited detection effect of low-resolution images, so as to increase stability and improve accuracy , reducing the effect of the collapse solution

Pending Publication Date: 2022-07-08
PING AN TECH (SHENZHEN) CO LTD
View PDF0 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] However, the detection effect of the two-dimensional key point detection method based on deep learning is limited for low-resolution images (specifically, the resolution is less than or equal to 72×300), because the low-resolution images reflect the connection relationship and positioning There is a disadvantage of insufficient natural information in tasks such as facet joints, so there is an urgent need for a human body key point detection method suitable for low-resolution images

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 key point detection method and system, electronic equipment and readable storage medium
  • Human body key point detection method and system, electronic equipment and readable storage medium
  • Human body key point detection method and system, electronic equipment and readable storage medium

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0047] see figure 1 , a human body key point detection method according to an embodiment of the present invention, specifically a low-resolution two-dimensional human body key point detection method based on contrastive learning, comprising the following steps:

[0048] Obtain an image to be detected; wherein, the image to be detected includes a human body image;

[0049] Input the image to be detected into the pre-trained two-dimensional human body key point detection model, and output the two-dimensional human body key point detection result;

[0050] Wherein, the step of obtaining the pre-trained two-dimensional human body key point detection model includes:

[0051] Process each original image in the pre-acquired unlabeled training data set with reduced resolution as a data enhancement method to obtain a corresponding processed image with reduced resolution data enhancement;

[0052] Each original image in the pre-acquired unlabeled training dataset and its corresponding...

Embodiment 2

[0059] The embodiment of the present invention may be based on the technical solution disclosed in Embodiment 1, and the twin network selects a twin CNN network with shared parameters. In another exemplary option, when the resolution is reduced as the data enhancement method, a bilinear interpolation algorithm is used.

[0060] In the embodiment of the present invention, for the detection of low-resolution human body key points, the data enhancement method of comparative learning is set to reduce the resolution, so as to adapt to the actual application scenario of low-resolution human body key point detection, so as to obtain a high-level image capable of extracting low-resolution images. Semantic feature feature extractor; using the low-resolution image high-level semantic feature extractor, fine-tuned for specific low-resolution human key point detection tasks, so that two-dimensional human key point detection can not only extract high-level semantic features of low-resolutio...

Embodiment 3

[0073] see Figure 4 , the embodiment of the present invention further explains the implementation of the system in combination with a low-resolution two-dimensional human body key point detection example using CNN as the encoder network application comparison learning; 1 ,x 2 ,x 3 ,...,x N } (such as LSP, FLIC.MPII single full body image dataset), labeled pedestrian image dataset (such as the MSCOCO dataset).

[0074] A. Model pre-training stage: Use a simple CNN model to perform comparative learning pre-training on dataset X. In this process, reducing the resolution is selected as the data enhancement method, and an image with a reduced resolution of the original picture is obtained. The original image and the low-resolution image are input into the twin CNN network with shared parameters, and the output vectors are obtained respectively. After one of the output vectors is output by the multi-layer perceptron, the negative cosine is performed between the other output v...

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 key point detection method and system, electronic equipment and a readable storage medium. The method comprises the following steps: acquiring a to-be-detected image; wherein the to-be-detected image comprises a human body image; and inputting the to-be-detected image into a pre-trained two-dimensional human body key point detection model, and outputting and obtaining a two-dimensional human body key point detection result. According to the method, the high-level semantic feature extractor of the low-resolution picture is obtained by utilizing comparative learning and label-free data, so that the dependence of manual annotation can be reduced; an asymmetric multilayer perceptron design and a gradient stopping strategy are utilized, so that the stability of the training process is improved, and collapse is reduced; the accuracy of low-resolution two-dimensional human body key point detection can be improved.

Description

technical field [0001] The invention belongs to the technical field of computer vision, relates to the field of human body posture estimation, and in particular relates to a human body key point detection method, system, electronic device and readable storage medium. Background technique [0002] Human Keypoints Detection, also known as human pose estimation, is a pre-task of human action recognition, behavior analysis, human-computer interaction, etc. in computer vision; among them, the two-dimensional human keypoint detection problem is defined as a single person Two-dimensional localization of human joint key points in whole body images, the main challenges of two-dimensional human key point detection are: 1) Capture the strong connection between human joint points; 2) There are some small joint points that are almost invisible 3) Ambient occlusion. [0003] Existing mainstream methods are mainly driven by the above-mentioned challenge 1), that is, the need to find all p...

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): G06V40/10G06V40/20G06V10/774G06V10/80G06V10/82G06K9/62G06N3/02
CPCG06N3/02G06F18/214G06F18/253
Inventor 司世景王健宗吴建汉
Owner PING AN TECH (SHENZHEN) CO LTD
Features
  • R&D
  • Intellectual Property
  • Life Sciences
  • Materials
  • Tech Scout
Why Patsnap Eureka
  • Unparalleled Data Quality
  • Higher Quality Content
  • 60% Fewer Hallucinations
Social media
Patsnap Eureka Blog
Learn More

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

© 2025 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com