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

Crowd counting method and system based on switching convolutional network

A crowd counting, convolutional network technology, applied in the field of image processing, can solve problems such as difficult to deal with crowd occlusion, and achieve high accuracy and robustness.

Pending Publication Date: 2020-03-17
SHANDONG NORMAL UNIV
View PDF3 Cites 4 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

Regardless of the detection-based method, it is difficult to deal with the severe occlusion problem between crowds

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 switching convolutional network
  • Crowd counting method and system based on switching convolutional network
  • Crowd counting method and system based on switching convolutional network

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0032] Such as figure 1 As shown, the present disclosure provides a crowd counting method based on switched convolutional networks, including:

[0033] S1. Divide the input image into blocks, and cut it into 9 non-overlapping parts, each of which is 1 / 3 of the length and width of the original image. Its purpose is to make the input image block can be regarded as having a single density, scale and perspective information, as a minimum unit for selecting a regressor;

[0034] S2. The image block is assigned a label to the input image block through the Switch-CNN classifier for classification, and a suitable CNN regressor is selected for density estimation;

[0035] The Switch-CNN classifier is a three-class classifier based on VGG16. The fully connected layer in VGG16 is removed, and the global average pooling (GAP) on Conv 5 features is used to remove spatial information and aggregate discriminative features. After the GAP is a smaller fully connected layer and a three-level ...

Embodiment 2

[0051] The present disclosure provides a crowd counting system based on switched convolutional networks, including:

[0052] A classification module, which is used to divide the target image into blocks, input the image block into the switching convolutional neural network, and classify it through the classifier according to the density;

[0053] A feature splicing module, which is used to extract the feature of the density map from the classified image block through the regressor, and obtain the feature map combined with the global density map feature through feature splicing for the obtained density map feature;

[0054] The calculation module is used to process the feature map combined with the global density feature through mean pooling and deconvolution layer processing to obtain the target estimated density map, and obtain the number of people in the target image by integrating the target estimated density map.

Embodiment 3

[0056] The present disclosure provides an electronic device, including a memory, a processor, and a computer program stored on the memory and operable on the processor, wherein the processor implements the switching based on Steps of a crowd counting method for convolutional 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 invention discloses a crowd counting method and system based on a switching convolutional network. The crowd counting method comprises the steps: partitioning a target image, inputting image blocks into the switching convolutional neural network, and carrying out the classification of the image blocks through a classifier according to the density; performing density map feature extraction on the classified image blocks through a regression device, and performing feature splicing on the obtained density map features to obtain a feature map combined with global density map features; and performing mean pooling and deconvolution layer processing on the feature map combined with the global density features to obtain a target estimated density map, and obtaining the number of people in thetarget image through integration. The switching convolution network takes a plurality of CNNs with different convolution kernel sizes as regression devices for density map prediction, uses a trained selection classifier to select an optimal CNN regression device for each input image, and takes the result as a final result, so that the accuracy and robustness of predicting the number of crowds areimproved.

Description

technical field [0001] The present disclosure relates to the technical field of image processing, in particular to a crowd counting method and system based on a switched convolutional 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] Crowd counting is a branch of crowd analysis that is essential for video surveillance, public area planning, traffic flow monitoring, etc. Its purpose is to be able to accurately predict the number of targets. Furthermore, the method used for swarm technology can be extended to counting tasks in other fields, such as cell microscopy, vehicle counting, environmental surveys, etc. However, the accuracy of the results is far from optimal due to various complexities such as occlusions, high clutter, uneven distribution of people, uneven lighting, variations in appearance, scale, and perspective within an...

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): G06K9/00G06K9/62G06N3/04G06N3/08
CPCG06N3/08G06V20/53G06N3/045G06F18/241
Inventor 吕蕾顾玲玉陈梓铭吕晨张桂娟刘弘
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