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

Low illumination target detection method based on rpf-cam

A technology of RPF-CAM and target detection, applied in neural learning methods, instruments, biological neural network models, etc., can solve problems such as poor detection of small objects, poor detection of low-light targets, multiple low-level information, etc.

Active Publication Date: 2022-04-05
GUILIN UNIV OF ELECTRONIC TECH
View PDF12 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] The current target detection network, such as the Faster R-CNN network, gradually down-samples during the feature extraction process, and finally sends the obtained feature map to the region proposal network to generate a priori frame, so that the obtained feature map has more The loss of low-level information leads to poor detection of small objects, and for low-illumination images, there is no targeted separation of illumination information and color information, resulting in poor detection of low-illumination objects

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
  • Low illumination target detection method based on rpf-cam
  • Low illumination target detection method based on rpf-cam
  • Low illumination target detection method based on rpf-cam

Examples

Experimental program
Comparison scheme
Effect test

Embodiment

[0087] A low-illuminance target detection method based on RPF-CAM, comprising the steps of:

[0088] 1) Create a synthetic data source and create a source image: integrate Nor-images of normal illumination images obtained under normal sunlight, Dark-images of low illumination images obtained by simulating the imaging characteristics of low illumination environments, and image annotation data Images_Annotation to construct A Dark-Nor-Data dataset, the grouping of the dataset is shown in Table 1 below:

[0089] Table 1:

[0090]

[0091]

[0092] 2) The training of the feature extraction network: including:

[0093] 2-1) Preprocess all low-illuminance images Dark-images and normal-illuminance images Nor-images, and scale them to a uniform width and height;

[0094] 2-2) Use the Lab color model to decompose the low-illuminance images Dark-images and normal-illuminance images Nor-images into two parts: light components and color components, respectively down-sample the two...

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 low-illuminance target detection method based on RPF-CAM, which is characterized in that it includes the following steps: 1) making a synthetic data source and establishing a source image; 2) training of a feature extraction network module; 3) channel attention The training of the mechanism network module; 4) the training of the target detection network; 5) the construction and training of the entire convolutional neural network. This method makes the feature expression ability of the feature map of target detection better and the accuracy of target detection is high.

Description

technical field [0001] The invention relates to the technical field of image enhancement and target detection, in particular to a low-illuminance target detection method based on residual pyramid fusion and channel attention mechanism RPF-CAM (residual pyramid fusion and channel attention mechanism, RPF-CAM for short). Background technique [0002] Existing target detection methods are mainly aimed at target detection in environments with normal illumination, but for low illumination, the image of the target to be detected is dark, blurred, and there are many interferences, the detection effect and detection accuracy are generally poor. less favorable situation. [0003] The Lab color model is a color model announced by the International Commission on Illumination (CIE) in 1976. It is a color model determined by the CIE organization that theoretically includes all colors visible to the human eye. The Lab model makes up for the two colors of RGB and CMYK. Insufficient patter...

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): G06V10/764G06V10/82G06K9/62G06N3/04G06N3/08
CPCG06N3/08G06N3/047G06N3/045G06F18/241G06F18/2415
Inventor 江泽涛翟丰硕
Owner GUILIN UNIV OF ELECTRONIC TECH
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