Bilateral attention mechanism-based target detection method under complex background

A technology of target detection and complex background, which is applied in the field of target detection under complex background in computer vision, can solve the problems of insufficient detection accuracy of target detection algorithms, achieve the effect of enhancing supervision and constraint capabilities, accurate detection results, and improving efficiency

Pending Publication Date: 2022-06-28
SOUTHEAST UNIV
View PDF0 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] Aiming at the problem of insufficient detection accuracy of existing target detection algorithms in complex backgrounds, the present invention proposes a target detection method in complex backgrounds based on a bilateral attention mechanism based on the current algorithm SINet, which is named Bi- SINet

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
  • Bilateral attention mechanism-based target detection method under complex background
  • Bilateral attention mechanism-based target detection method under complex background
  • Bilateral attention mechanism-based target detection method under complex background

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0049] see Figure 1-Figure 5 , this implementation provides a method for detecting objects in complex backgrounds based on a bilateral attention mechanism.

[0050] Specifically, see figure 1 , this method specifically includes:

[0051] Step S1: Acquire public target detection data sets under complex backgrounds, including: COD10K data sets, CAMO data sets and CHAMELEON data sets, and construct training sets, validation sets and test sets accordingly;

[0052] More specifically, the constructed training set contains 3040 pairs of image data in the COD10K dataset and 100 pairs of image data in the CAMO pair dataset, with a total of 4040 pieces of data; the constructed validation set contains 101 pairs of images in the COD10K dataset. Data pairs; the constructed test set contains 2026 pairs of image data in COD10K, 250 pairs of image data in CAMO dataset and 76 pairs of image data in CHAMELEON dataset, with a total of 2352 pieces of data.

[0053] Step S2: Construct a Bi-SI...

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 bilateral attention mechanism-based target detection method under a complex background, which can be used for carrying out accurate foreground target detection under the complex background. The method mainly comprises the following steps: constructing a training set, a verification set and a test set according to a disclosed target detection data set under a complex background; constructing an artificial neural network detection model Bi-SINet based on a bilateral attention mechanism; an SGD optimizer is used on a Pytorch deep learning platform to optimize the Bi-SINet model; and the detection performance of the convergent Bi-SINet network model is evaluated on the constructed test set. Compared with a current main target detection algorithm SINet under a complex background, the method provided by the invention can obtain better detection performance. According to the method, the average absolute error is reduced, a higher enhancement-alignment index, a structure index and a weighted F index are realized, and the method is a more accurate target detection algorithm under a complex background.

Description

technical field [0001] The invention relates to a target detection method under complex background based on bilateral attention mechanism, which is applicable to the technical field of target detection under complex background in computer vision. Background technique [0002] Images and videos are an important source of information for human beings, so the application of computer analysis and processing for massive images and videos has also been vigorously developed. As one of the basic tasks in the field of computer vision, object detection is an important means to help computers understand image data. It has a wide range of application prospects in pedestrian detection, vehicle detection, automatic driving, security systems and medical care. [0003] Thanks to the development of deep learning technology, object detection has attracted extensive attention in recent years and achieved great success, and a large number of efficient detection algorithms have been proposed. E...

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): G06V10/80G06V10/82G06V10/774G06V10/776G06V10/26G06N3/04G06N3/08G06K9/62
CPCG06N3/08G06N3/045G06F18/2193G06F18/253G06F18/214
Inventor 李春国罗顺刘周勇杨绿溪
Owner SOUTHEAST UNIV
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap