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

Crowd counting method and system based on scale adaptive network

A scale-adaptive and crowd counting technology, applied in the field of image processing, can solve problems such as complex backgrounds, little application significance, and head size differences, and achieve the effects of alleviating negative impacts, avoiding complex tasks, and increasing the range

Active Publication Date: 2020-12-08
SHANDONG UNIV
View PDF5 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, crowd counting still faces great challenges due to problems such as differences in head size, irregular distribution of crowds, and complex backgrounds.
[0004] The inventor found that most of the existing crowd counting methods focus on outputting a single number to represent the number of people, and cannot display detailed information such as crowd distribution, so the practical application is of little significance
Since 2015, a crowd counting method that outputs a density map and obtains the number of people based on the density map has gradually appeared, but the ability to deal with multi-scale targets and complex backgrounds is weak, and the calculation time is relatively long

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
  • Crowd counting method and system based on scale adaptive network
  • Crowd counting method and system based on scale adaptive network
  • Crowd counting method and system based on scale adaptive network

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0037] In view of the wide application of deep learning in the field of machine vision (tracking, detection, positioning, etc.) and the powerful performance of convolutional neural networks in image processing and feature learning.

[0038] In one or more embodiments, a method for counting people based on a scale-adaptive network is disclosed. The method aims at issues such as head size differences and complex backgrounds, using a combination of dilated convolutional neural networks and traditional convolutional neural networks. Get a feature output density map with multiple receptive fields and perform crowd counting:

[0039] The traditional convolutional neural network and the expanded convolutional neural network are used for crowd counting; in order to obtain features with more receptive fields to deal with head size differences and complex backgrounds, multiple units are densely linked; in order to reduce multiple feature channels Competitiveness among them, a channel at...

Embodiment 2

[0061] In one or more implementations, a scale-adaptive network-based people counting system is disclosed, including:

[0062] A module for obtaining the original image containing crowds and performing scaling processing on the original image;

[0063] A module for generating a corresponding density map according to the number of people label of the sample; the number of people label of the sample refers to the position of the head center marked in the original image in the image;

[0064] A module for intercepting a set number of image blocks from a scaled image, a module for intercepting a set number of density image blocks from a density map;

[0065] A module for building scale-adaptive crowd counting networks based on dilated convolutional neural networks and channel attention mechanisms;

[0066] means for training a scale-adaptive crowd counting network using said image patches and density image patches;

[0067] A module for calculating the density map of each image ...

Embodiment 3

[0069] In one or more embodiments, a terminal device is disclosed, including a server, the server includes a memory, a processor, and a computer program stored on the memory and operable on the processor, and the processor executes the The program implements the crowd counting method based on the expanded convolutional neural network in the first embodiment. For the sake of brevity, details are not repeated here.

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 method and system for counting the number of people based on a scale-adaptive network. block; use the image block and the corresponding density block of each image block to train the scale adaptive network; use the trained scale adaptive network to output the density map for each image, and sum all the pixels in the density map , and finally get all the people in the original image. The invention effectively improves the accuracy of crowd counting and the robustness to head size differences and complex backgrounds.

Description

technical field [0001] The invention belongs to the field of image processing, and in particular relates to a method and system for counting the number of people of a crowd based on a scale adaptive network. Background technique [0002] The statements in this section merely provide background information related to the present invention and do not necessarily constitute prior art. [0003] Crowd Counting refers to counting the number of individual targets for the crowd in the video or image. In recent years, crowd counting based on pattern recognition and machine learning has been widely researched and applied in the field of intelligent monitoring, such as: the monitoring of the flow of people in airports and stations and the distribution of crowds in large shopping malls. By monitoring the number of people in a certain place, real-time density information can be provided to the management agency, and the flow of people can be effectively controlled, thereby providing man...

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): G06K9/00G06T3/40G06T7/77
CPCG06T3/40G06T7/77G06T2207/20081G06T2207/20084G06T2207/30196G06T2207/30242G06V20/53
Inventor 常发亮张友梅李南君
Owner SHANDONG 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