An unsupervised image inpainting method based on mask generation against network migration learning

A technology of transfer learning and inpainting method, applied in the field of unsupervised image inpainting based on mask generation confrontation network transfer learning, which can solve problems such as unsupervised image inpainting and inapplicability of algorithms, achieve excellent performance, improve visual performance, and improve quality Effect

Active Publication Date: 2019-02-05
聚时科技(上海)有限公司
View PDF7 Cites 8 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] The second category is unsupervised image inpainting
However, the full image may not always be available in practical applications, making these algorithms unsuitable for

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
  • An unsupervised image inpainting method based on mask generation against network migration learning
  • An unsupervised image inpainting method based on mask generation against network migration learning
  • An unsupervised image inpainting method based on mask generation against network migration learning

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0043] The present invention will be described in detail below in conjunction with the accompanying drawings and specific embodiments. This embodiment is carried out on the premise of the technical solution of the present invention, and detailed implementation and specific operation process are given, but the protection scope of the present invention is not limited to the following embodiments.

[0044] 1. GAN

[0045] Generative adversarial networks have achieved great success in generating realistic images. It mainly consists of two parts, a generator (Gen) and a discriminator (Disc). The loss function is optimized based on max-min game theory to achieve a balance between Gen and Disc, and the training process consists of two steps. Disc is trained to distinguish between synthetic images generated by Gen and real images sampled from real image datasets. Gen, on the other hand, is trained to generate fake images that can be confused with real ones. The goal of training is...

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 relates to an unsupervised image restoration method for resisting network migration learning based on mask generation, comprising the following steps: step S101, obtaining an image to berepaired and setting mask parameters;step 102, utilize a mask to generate an antagonistic network for repairing that image to be repaired; The mask generation countermeasure network comprises a self-encoder and a discriminator. The training process of the mask generation countermeasure network comprises adding a mask layer after the output of the self-encoder, and the mask layer is used as the first layer of the discriminator to realize the training of the self-encoder and the discriminator. Compared with the prior art, the invention has the advantages of high accuracy and high efficiency.

Description

technical field [0001] The invention relates to an image restoration method based on artificial intelligence, in particular to an unsupervised image restoration method based on mask generation confrontation network migration learning. Background technique [0002] In practical applications, images sometimes have broken or blurred areas that need to be repaired. The goal of image restoration is to reconstruct the missing or corrupted parts of an image. Over the past few years, image inpainting has been intensively studied, focusing on reconstructing retouched or incomplete parts of an image. These studies usually employ autoencoders as the basic structure and incorporate other tricks, such as adversarial learning, to improve performance. According to whether the full image is known or not, previous works can be mainly divided into two categories. [0003] The first category is supervised image inpainting. Supervised image inpainting methods focus on problems known from th...

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): G06T5/00G06N3/08
CPCG06N3/084G06N3/088G06T5/001
Inventor 郑军刘新旺
Owner 聚时科技(上海)有限公司
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