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

Infrared target detection method based on improved YOLOv3

A detection method, infrared target technology, applied in the field of computer vision, can solve the problem of missing feature information, achieve the effect of increasing network depth, weakening the effect of image feature loss, and avoiding information loss

Active Publication Date: 2021-02-09
中国人民解放军火箭军工程大学
View PDF4 Cites 2 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

Another deep learning network commonly used for target detection is the residual network, which has a good performance in processing image classification tasks, but the rapid change in the number of channels of the network feature map will lead to the loss of feature information

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
  • Infrared target detection method based on improved YOLOv3
  • Infrared target detection method based on improved YOLOv3
  • Infrared target detection method based on improved YOLOv3

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0035] Based on the improved YOLOv3 for infrared target detection, Darknet-53 is used as the network detection framework. Darknet-53 consists of 53 convolutional layers. refer to figure 1 , the basic structure of Darknet-53 is inside the dotted line box. Firstly, the last convolutional layer of Darknet-53 is removed. A residual module is added after the convolutional layer of the repeated block, and the low-level features are fused using the residual layer.

[0036] refer to figure 2 with image 3 , adding an attention module at the bottom of the repeating block, consisting of a backbone branch and a mask branch. The main branch performs convolution and other processing on the features to obtain T(x), and the mask branch performs two downsampling and two upsampling on the feature map to obtain M(x) of the same size, and the point multiplication between the mask branch and the main branch The result, as the output result, can be expressed as

[0037] H(x)=M(x)·T(x)

[00...

Embodiment 2

[0050] The difference between this embodiment and Embodiment 1 is that the number of channels of the repeated block in the network detection framework increases with the number of repetitions, and the number of channels of the repeated block repeated for the nth time can be expressed as

[0051]

[0052] where D' start Input the number of feature map channels of the current repeated block, D′ end is the number of feature map channels output by the current repeated block, and λ is the repeated coefficient of the current repeated block.

[0053] Other methods and steps of this embodiment are the same as those in Embodiment 1.

[0054] Case Analysis

[0055] Two sets of experimental data were collected.

[0056] The experimental data of group A adopts the method of self-made infrared data set, and the infrared system is used for data collection in the laboratory, and the N-scale model is taken as the data source, and special vehicles, ships, and aircraft are used as the res...

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 an infrared target detection method based on improved YOLOv3, and the method comprises the steps: taking Darknet-53 as a network detection framework, removing a convolution layer between the network detection framework and a prediction module, adding multi-scale fusion prediction, and fusing repeated blocks of low-level features through a residual layer; adding attention modules to the bottoms of the repeated blocks, and adding a residual pyramid transition network between the repeated blocks, wherein the number of channels of the repeated blocks is increased progressively along with the number of times of repetition. The method has the advantages of being good in feature extraction capacity and information transition level when the target of the infrared image is detected.

Description

technical field [0001] The invention belongs to the technical field of computer vision, and in particular relates to an infrared target detection method based on improved YOLOv3. Background technique [0002] Whether infrared terminal guidance can quickly and accurately find and identify targets is extremely important for missile weapons. Usually, the collected infrared image weakens or even discards the texture and color of the object, and highlights its shape, contour and heat distribution. Traditional methods mainly use image enhancement and feature extraction to detect infrared targets. With the development of science and technology, infrared target detection technology has gradually migrated from traditional methods to deep learning methods, which has greatly improved the accuracy and speed of detection compared with the past. Infrared target detection based on deep learning has strong feature extraction and learning generalization capabilities, and is currently a hot...

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): G06K9/62
CPCG06F18/241
Inventor 付光远陈铁明魏振华汪洪桥张少磊
Owner 中国人民解放军火箭军工程大学
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