Check patentability & draft patents in minutes with Patsnap Eureka AI!

Multi-target detection method based on convolutional neural network

A convolutional neural network and detection method technology, applied in neural learning methods, biological neural network models, neural architectures, etc., can solve problems such as insufficient detection accuracy of small targets, and achieve improved accuracy, improved algorithm accuracy, and good fusion features. Effect

Pending Publication Date: 2021-06-04
XIDIAN UNIV
View PDF9 Cites 6 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0005] The purpose of the present invention is to address the deficiencies of the above-mentioned prior art, and propose a multi-target detection method based on convolutional neural network, which is used to solve the technical problem of insufficient detection accuracy of small targets in existing real-time target detection methods

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
  • Multi-target detection method based on convolutional neural network
  • Multi-target detection method based on convolutional neural network
  • Multi-target detection method based on convolutional neural network

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0026] Refer to attached figure 1 , a kind of multi-target detection method based on convolutional neural network that the present invention proposes, comprises the following steps:

[0027] Step S1: Acquiring image data of the target to be detected;

[0028] In the embodiment of the present invention, the multi-target detection method is applied to the industrial camera interface image acquisition platform, and the industrial camera is used to collect image data to realize multi-target detection, which makes the application range and environment more extensive.

[0029] Step S2: Extract image data to obtain multi-layer feature maps.

[0030] In the embodiment of the present invention, the yolov5 detection framework is selected as the improved benchmark model. After the industrial camera acquires the image data, the image data is spliced ​​by random zooming, random cutting, random arrangement, etc., to enrich the detection data set.

[0031] Using the backbone network in the...

Embodiment 2

[0037] Based on step S3 of Embodiment 1, the embodiment of the present invention provides a gated spatial pyramid cavity convolutional network, the structure of which is as follows figure 2 As shown, it includes: input layer 101, gating mechanism 102, first convolution 103, second convolution 104, third convolution 105, fourth convolution 106, connection unit 107, fifth convolution 108 and output layer 109.

[0038] The input layer 101 inputs the feature maps into the gating mechanism 102, the first convolution 103, the second convolution 104, the third convolution 105 and the fourth convolution 106, respectively. The outputs of the four convolutions are respectively multiplied by the output of the gating mechanism 102 , and then the multiplication results are connected through the connection unit 107 . The purpose of the fifth convolution 108 is to adjust the number of output channels, so that the output result of the connection unit 107 outputs the first fusion feature map...

Embodiment 3

[0041] Based on step S3 of embodiment 1, after obtaining the second fusion feature map, the present invention introduces an attention mechanism; refer to Figure 4 , the attention mechanism network includes: a second global pooling layer 301 , a seventh convolution 302 , a second activation function 303 , an eighth convolution 304 and a third activation function 305 . The input data undergoes global average pooling through the second global pooling layer 301, and then performs channel compression through the seventh convolution 302, activates with the second activation function 303, restores the number of channels through the eighth convolution 304, and finally uses the third activation Function 305 generates final channel weights and outputs the result. In this embodiment, the seventh convolution 302 and the eighth convolution 304 are pointwise convolutions with a size of 1×1, the second activation function uses the Hardswish function, and the third activation function uses t...

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 multi-target detection method based on a convolutional neural network. The technical problem that an existing real-time target detection method is insufficient in small target detection precision is mainly solved. The implementation scheme comprises the following steps: acquiring image data of a to-be-detected target; analyzing the image data through a convolutional neural network to obtain a multi-layer feature map; fusing the multi-layer feature maps through a feature fusion network, learning importance degrees of different receptive field branches, and fusing and outputting high-layer global semantic information and bottom-layer local detail information to obtain a third fusion feature map; and finally, generating a candidate frame from the third fusion feature map according to a preset scale, and analyzing and processing the candidate frame to obtain a target detection result, thereby realizing multi-target detection. According to the method, the information correlation between the features is enhanced by fusing the information of the multi-layer feature map, so that the accuracy of multi-target detection can be effectively improved.

Description

technical field [0001] The invention relates to the technical field of artificial intelligence, in particular to a multi-target detection method based on a convolutional neural network. Background technique [0002] Since the development of deep learning, computer vision has gradually become a very popular research direction at home and abroad. The so-called computer vision is to let the computer have human-like vision, be able to "see" the information of the outside world, and have the ability of human beings to process information, including tasks such as image classification, target detection, target tracking and image segmentation, among which target detection is The basic link plays a vital role in subsequent tasks such as target tracking. The task of multi-object detection is to mark all the objects of interest in an image with bounding boxes and obtain the category information of the objects. [0003] In recent years, with the rapid development and popularization of...

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): G06K9/46G06K9/62G06N3/04G06N3/08
CPCG06N3/08G06V10/449G06N3/048G06N3/045G06F18/241Y02P90/30
Inventor 肖嵩张兆琦杨子轩杨翌晗张同振董文倩曲家慧
Owner XIDIAN UNIV
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