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

An MCNN crowd counting method based on global density characteristics

A technology of crowd counting and density, which is applied in the field of crowd image processing, can solve the problems of loss of detail features, not fully considering the global density changes of crowd images, etc., and achieve high accuracy and robustness

Inactive Publication Date: 2019-06-28
ZHEJIANG UNIV OF TECH
View PDF1 Cites 11 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, the existing crowd counting methods based on deep learning do not fully consider the global density changes in the crowd image, and the convolutional neural network generally has a pooling layer, and the detailed features will be lost during the downsampling process of the pooling layer.

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
  • An MCNN crowd counting method based on global density characteristics
  • An MCNN crowd counting method based on global density characteristics
  • An MCNN crowd counting method based on global density characteristics

Examples

Experimental program
Comparison scheme
Effect test

Embodiment

[0038] Example: such as figure 1 As shown, the MCNN crowd counting method based on the combination of global density features includes the following steps:

[0039] Step 1: Use the two-dimensional Gaussian convolution kernel to convert the crowd image into a crowd density map label according to the manually marked head coordinates in the data set. An image label with N heads can be represented by the following formula (1):

[0040]

[0041] Among them, G σ (x) is a two-dimensional Gaussian convolution kernel, σ is its width parameter, δ(x-x i) is the delta function, x i Indicates the location of a human head marker point.

[0042] And, according to the real number of people in the image in the data set, the density level label of each image is generated. After reading the data set, the maximum number of people in the data set N max with minimum number N min Subtract to determine the variation range of the number of people, and specify the number M of density levels to ...

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 an MCNN crowd counting method based on global density characteristics. The crowd counting method can adapt to different density scenes, is capable of generating the crowd density map containing more comprehensive information, obtains the estimated number of people by integrating the crowd density map, and can avoid the influence of crowd occlusion and non-uniform crowd distribution in the crowd image on counting; and the convolutional neural network combined with global density characteristics is utilized to estimate the population density map, so that the method can be suitable for a non-uniform population distribution scene, the density map containing more comprehensive information is generated, and the population estimation result has higher accuracy and robustness.

Description

technical field [0001] The invention relates to the field of crowd image processing, in particular to an MCNN crowd counting method based on combining global density features. Background technique [0002] With the development of social economy, the scale of urban population is increasing, and the number of people gathered in public places such as stations, squares, and parks is also increasing. Large-scale crowd gathering may lead to safety accidents such as trampling. In order to better ensure personal safety, the research on crowd counting algorithms is very important. [0003] Existing crowd counting methods can be divided into three categories: direct counting methods based on object detection, indirect counting methods based on feature regression, and crowd counting methods based on deep learning. Among them, the direct counting method based on target detection uses the number of pedestrians detected from the image to count. This type of method is suitable for scenes ...

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): G06K9/00G06N3/04G06N3/08
Inventor 刘志陈越沈国江
Owner ZHEJIANG 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

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