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

A crowd counting method and system based on vertical and horizontal cross attention network

A criss-cross and crowd counting technology, applied in computing, computer parts, instruments, etc., can solve the problems of low accuracy of counting results, low computing efficiency, and large resource consumption, so as to improve efficiency, resource consumption, and improve efficiency and accuracy, the effect of reducing misestimation

Active Publication Date: 2022-07-22
SHANDONG NORMAL UNIV
View PDF5 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

Detection-based crowd counting has high detection accuracy in sparse scenes, but in dense scenes, especially in the presence of occlusions and background clutter, its results will be unsatisfactory; regression-based counting successfully solves the However, the current mainstream method is based on convolutional neural network crowd counting. This method mainly includes two network structures, single column and multi-column. A single column is generally deployed with a single and deep depth. Convolutional neural network, but it ignores the multi-scale information in the scene, making the accuracy of the counting results low; multiple columns generally use different columns to capture the multi-scale information in the scene, but the multi-column structure often has many parameters, Bloated network will consume a lot of resources, and the calculation efficiency is low

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
  • A crowd counting method and system based on vertical and horizontal cross attention network
  • A crowd counting method and system based on vertical and horizontal cross attention network
  • A crowd counting method and system based on vertical and horizontal cross attention network

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0039] In this example, as figure 1 As shown, a crowd counting method based on the vertical and horizontal cross attention network is disclosed, including:

[0040] Get crowd images;

[0041] Extract local feature maps from crowd images;

[0042] Input the local feature map into the cyclic vertical and horizontal cross attention module, and output the attention feature map;

[0043] Obtain the crowd density map according to the attention feature map;

[0044] According to the crowd density map, the number of crowds corresponding to the crowd image is obtained.

[0045] Further, the crowd images are preprocessed, and local feature maps are extracted from the preprocessed crowd images.

[0046] Further, the crowd image is input into the local feature extraction module to extract the local feature map.

[0047] Further, the local feature extraction module includes the convolution layer of VGG-16Net and the dilated convolution module. After the initial feature map is extracte...

Embodiment 2

[0069] In this embodiment, a crowd counting system based on a vertical and horizontal cross attention network is disclosed, including:

[0070] Image acquisition module for acquiring crowd images;

[0071] The local feature map acquisition module is used to extract local feature maps from crowd images;

[0072] The attention feature map acquisition module is used to input the local feature map into the circular vertical and horizontal cross attention module, and output the attention feature map;

[0073] The crowd density map acquisition module is used to obtain the crowd density map according to the attention feature map;

[0074] The crowd counting module is used to obtain the crowd number corresponding to the crowd image according to the crowd density map.

Embodiment 3

[0076] In this embodiment, an electronic device is disclosed, which includes a memory, a processor, and computer instructions stored in the memory and executed on the processor. When the computer instructions are executed by the processor, one of the methods disclosed in Embodiment 1 is completed. A Crowd Counting Method Based on Cross-Attention Networks.

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 present disclosure discloses a crowd counting method and system based on a vertical and horizontal cross attention network, including: acquiring a crowd image; extracting a local feature map from the crowd image; inputting the local feature map into a circular vertical and horizontal cross attention module, and outputting attention feature map; obtain the crowd density map according to the attention feature map; obtain the number of crowds corresponding to the crowd image according to the crowd density map. The contextual information of the image is obtained through the circular criss-cross attention module, which improves the efficiency and accuracy of crowd counting.

Description

technical field [0001] The invention relates to the technical field of crowd counting, in particular to a crowd counting method and system based on a vertical and horizontal cross attention network. Background technique [0002] The statements in this section merely provide background information related to the present disclosure and do not necessarily constitute prior art. [0003] With the substantial increase in population density in cities, people gather more and more. For the purpose of crowd control and public safety, accurately estimating the number of people from images or videos has become an important application in computers. Due to the influence of factors such as scale change, occlusion, uneven crowd distribution, and illumination change in the image, the algorithm of crowd counting is subject to certain challenges. [0004] There are three main methods of crowd counting: detection-based counting, regression-based counting and convolutional neural network-based...

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 Patents(China)
IPC IPC(8): G06V20/52G06V10/44G06V10/82
CPCG06V20/53G06V10/44
Inventor 康春萌孟琛盛星吕蕾
Owner SHANDONG NORMAL 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