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

Image amplification method and system based on generative adversarial network

An image and network technology, applied in the field of image processing, can solve problems such as disadvantage, inability to perform image augmentation operations, and scarcity of image training sets.

Inactive Publication Date: 2019-07-30
GUANGZHOU XIAOPENG MOTORS TECH CO LTD
View PDF25 Cites 10 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

In practice, it is found that traditional image augmentation methods cannot perform image augmentation operations based on the content of the target foreground area in the image, resulting in the scarcity of image training sets in the actual environment, which is not conducive to training the recognition system through image training sets to improve recognition The target recognition rate of the system

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
  • Image amplification method and system based on generative adversarial network
  • Image amplification method and system based on generative adversarial network
  • Image amplification method and system based on generative adversarial network

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0120] see figure 1 , figure 1 It is a schematic flowchart of an image augmentation method based on a generative confrontation network disclosed in an embodiment of the present invention. Such as figure 1 As shown, the image augmentation method based on the generative confrontation network may include the following steps:

[0121] 101. The image augmentation system performs normalization and preprocessing on the collected training images to obtain an initial data set of training images.

[0122] In the embodiment of the present invention, the image augmentation system can adopt the object-oriented design method and software engineering specification in Ubuntu 16.04 of the microcomputer, and implement the image based on the generative confrontation network disclosed in the embodiment of the present invention through Python computer voice. The amplification method, wherein, Ubuntu 16.04 means the 16.04 version of Ubuntu, Ubuntu or called Ubuntu, is an open source GNU / Linux op...

Embodiment 2

[0209] see Figure 5 , Figure 5 It is a schematic structural diagram of an image augmentation system based on a generative confrontation network disclosed in an embodiment of the present invention. Such as Figure 5 As shown, the image augmentation system may include:

[0210] A processing unit 501, configured to normalize and preprocess the collected training images to obtain an initial dataset of training images;

[0211] The elimination unit 502 is used to obtain the confidence degree of the target foreground region of the initial data set through the target region proposal network, and perform geometric shape binary elimination on the target foreground region according to the confidence degree, so as to obtain the WGAN model training set;

[0212] The training unit 503 is used to train the WGAN model according to the WGAN model training set to obtain the target WGAN model;

[0213] The augmentation unit 504 is configured to input the image to be augmented into the tar...

Embodiment 3

[0261] see Figure 6 , Figure 6 It is a schematic structural diagram of another image augmentation system based on a generative confrontation network disclosed in an embodiment of the present invention. Such as Figure 6 As shown, the image augmentation system based on the generative confrontation network may include:

[0262] A memory 601 storing executable program codes;

[0263] a processor 602 coupled to the memory 601;

[0264] Wherein, the processor 602 invokes the executable program code stored in the memory 601 to execute figure 1 A generative adversarial network-based approach to image augmentation is described.

[0265] The embodiment of the present invention discloses a computer-readable storage medium, which stores a computer program, wherein the computer program enables the computer to execute figure 1 A generative adversarial network-based approach to image augmentation is described.

[0266] The embodiment of the present invention also discloses an appli...

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 embodiment of the invention relates to the technical field of image processing, and discloses an image amplification method and system based on a generative adversarial network, and the method comprises the steps: carrying out the normalization and preprocessing of a collected training image, so as to obtain an initial data set of the training image; obtaining a confidence coefficient of a target foreground area of the initial data set through a target area proposal network, and performing geometric shape binary elimination on the target foreground area according to the confidence coefficient to obtain a WGAN model training set; training the WGAN model according to the WGAN model training set to obtain a target WGAN model; and inputting the to-be-amplified image into the target WGAN model to obtain a target amplified image set. By implementing the embodiment of the invention, the rare condition limitation of the image training set can be overcome, and the recognition system can betrained through the image training set to improve the target recognition rate of the recognition system.

Description

technical field [0001] The invention relates to the technical field of image processing, in particular to an image augmentation method and system based on a generative confrontation network. Background technique [0002] With the rapid development of machine vision theory and models, many recognition systems based on machine vision have come out one after another. The recognition system based on machine vision can replace manual target recognition or data acquisition in occasions where artificial vision is difficult to meet the requirements. Therefore, It is widely used in electronics, automobile manufacturing and other industries. [0003] In order to improve the target recognition rate of the recognition system, a large number of image training sets are usually required to train the recognition system, and a large number of image training sets are usually obtained by performing an amplification operation on the image training set. Traditional image augmentation methods ma...

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): G06K9/62G06K9/32G06N3/04G06N3/08
CPCG06N3/08G06V10/25G06V2201/07G06N3/045G06F18/2148
Inventor 程子耀单文龙
Owner GUANGZHOU XIAOPENG MOTORS TECH CO LTD
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