检索结果-南通市图书馆

Implicit pairs for boosting unpaired image-to-image translation

Visual Informatics 2020年第4期4卷 50-58页

作者： Yiftach Ginger Dov Danon Hadar Averbuch-Elor Daniel Cohen-Or Tel Aviv University Israel

In image-to-image translation the goal is to learn a mapping from one image domain to *** the case of supervised approaches the mapping is learned from paired ***,collecting large sets of image pairs is often either prohibitively expensive or not *** a result,in recent years more attention has been given to techniques that learn the mapping from unpaired *** our work,we show that injecting implicit pairs into unpaired sets strengthens the mapping between the two domains,improves the compatibility of their distributions,and leads to performance boosting of unsupervised techniques by up to 12%across several *** competence of the implicit pairs is further displayed with the use of pseudo-pairs,i.e.,paired samples which only approximate a real *** demonstrate the effect of the approximated implicit samples on image-to-image translation problems,where such pseudo-pairs may be synthesized in one direction,but not in the *** further show that pseudo-pairs are significantly more effective as implicit pairs in an unpaired setting,than directly using them explicitly in a paired setting.

关键词： Generative adversarial networks image-to-image translation Data augmentation Synthetic samples

来源：

维普期刊数据库评论

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Edge-guided Adversarial Network Based on Contrastive Learning for image-to-image translation

Edge-guided Adversarial Network Based on Contrastive Learnin...

引用

第40届中国控制会议

作者： Chen Zhu Ru Lai Luzheng Bi Xuyang Wang Jiarong Du School of Automation Beijing Institute of Technology School of Mechnical Engineering Beijing Institute of Technology

In recent years, generative adversarial networks have made great progress in image synthesis and image translation tasks in the field of image processing and computer vision. However, the quality of the generated image and the scalability over multiple datasets is still not satisfying. We briefly review some prior works and propose a method for image-to-image translation,which is learning a mapping between different visual domains. The network extracts edge feature from both domains of output and target, and minimizes the difference using a framework based on patchwise contrastive learning. We apply edge feature guidance in our method and select Sobel operator among several classical edge detection operators. We demonstrate that our method outperforms existing approaches in the task of unpaired image-to-image translation across datasets.

关键词： Generative adversarial network image-to-image translation Contrastive learning Edge detection

来源： cnki会议评论

在线全文

cnki会议

学校读者我要写书评

暂无评论

Quantitative Cerebral Blood Volume image Synthesis from Conventional MRI Using image-to-image translation Techniques in Brain Tumors

Quantitative Cerebral Blood Volume Image Synthesis from Conv...

引用

第十六届中国医师协会神经外科医师年会

作者： LIU Yingchao

Background:Cerebral blood volume(CBV) mapping is the most accepted and robust method to evaluate tumor perfusion conditions,but it is not a standard process and will be missing in some clinical *** study aimed to investigate the feasibility of image-to-image translation techniques in generating CBV maps from conventional MR images of brain tumors with the help of the Bookend perfusion ***:A total of 756 MRI scans,including quantitative CBV maps produced by Bookend perfusion,were included in this retrospective *** were randomly assigned to a training cohort and validation cohort at a ratio of 7:*** sense-consistency generative adversarial network(SGAN) and the 3D encoder-decoder with only mean absolute error loss(MAE-ED) were trained to synthesize the CBV *** evaluation(four metrics below) and qualitative evaluation(a subjective score that ranged from 0 to 3) were performed to evaluate the quality of the synthesized CBV ***:For the quantitative evaluations,MAE-ED using pre-and post-T1 weighted imaging and apparent diffusion coefficient maps achieved the highest performance,with a mean absolute error of 1.76,mean squared error of 0.34,structural similarity of 86.29 and peak signal-to-noise ratio of *** qualitative evaluations of synthesized CBV via MAE-ED,a weighted average value of 2.63 was reached,and the ratios of zero to one,one to two,and two to three were 3.6%,29.6% and 66.8%,***:With the help of the Bookend perfusion technique,the image-to-image translation technique could produce a qualified and convincing CBV map from conventional MR images for reliable clinical *** of the study:A dynamic susceptibility contrast perfusion-weighted imaging-derived cerebral blood volume(CBV) map is the most accepted and robust method to evaluate tumor perfusion *** synthesis of CBV maps has become urgent due to the lack of standardized CBV maps without interoperator divergenc

关键词： quantitative cerebral blood volume generative adversarial network image-to-image translation MRI brain tumors

来源： cnki会议评论

在线全文

cnki会议

学校读者我要写书评

暂无评论

image to image translation Based on Differential image Pix2Pix Model

引用

Computers, Materials & Continua 2023年第10期77卷 181-198页

作者： Xi Zhao Haizheng Yu Hong Bian College of Mathematics and System Sciences Xinjiang UniversityUrumqi830017China School of Mathematical Sciences Xinjiang Normal UniversityUrumqi830017China

In recent years,Pix2Pix,a model within the domain of GANs,has found widespread application in the field of image-to-image ***,traditional Pix2Pix models suffer from significant drawbacks in image generation,such as the loss of important information features during the encoding and decoding processes,as well as a lack of constraints during the training *** address these issues and improve the quality of Pix2Pixgenerated images,this paper introduces two key ***,to reduce information loss during encoding and decoding,we utilize the U-Net++network as the generator for the Pix2Pix model,incorporating denser skip-connection to minimize information ***,to enhance constraints during image generation,we introduce a specialized discriminator designed to distinguish differential images,further enhancing the quality of the generated *** conducted experiments on the facades dataset and the sketch portrait dataset from the Chinese University of Hong Kong to validate our proposed *** experimental results demonstrate that our improved Pix2Pix model significantly enhances image quality and outperforms other models in the selected ***,the Pix2Pix model incorporating the differential image discriminator exhibits the most substantial improvements across all *** analysis of the experimental results reveals that the use of the U-Net++generator effectively reduces information feature loss,while the Pix2Pix model incorporating the differential image discriminator enhances the supervision of the generator during *** of these enhancements collectively improve the quality of Pix2Pix-generated images.

关键词： image-to-image translation generative adversarial networks U-Net++ differential image Pix2Pix

来源：

维普期刊数据库评论

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Unsupervised image translation with distributional semantics awareness

引用

Computational Visual Media 2023年第3期9卷 619-631页

作者： Zhexi Peng He Wang Yanlin Weng Yin Yang Tianjia Shao State Key Lab of CAD&CG Zhejiang UniversityHangzhou 310058China School of Computing University of LeedsLeedsUK School of Computing Clemson UniversityClemsonUSA

Unsupervised image translation(UIT)studies the mapping between two image *** such mappings are under-constrained,existing research has pursued various desirable properties such as distributional matching or two-way *** this paper,we re-examine UIT from a new perspective:distributional semantics consistency,based on the observation that data variations contain semantics,e.g.,shoes varying in ***,the semantics can be multi-dimensional,e.g.,shoes also varying in style,functionality,*** two image domains,matching these semantic dimensions during UIT will produce mappings with explicable correspondences,which has not been investigated *** propose distributional semantics mapping(DSM),the first UIT method which explicitly matches semantics between two *** show that distributional semantics has been rarely considered within and beyond UIT,even though it is a common problem in deep *** evaluate DSM on several benchmark datasets,demonstrating its general ability to capture distributional *** comparisons show that DSM not only produces explicable mappings,but also improves image quality in general.

关键词： generative adversarial networks(GANs) manifold alignment unsupervised learning image-to-image translation distributional semantics

来源：

维普期刊数据库评论

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Review of GAN-Based Research on Chinese Character Font Generation

引用

Chinese Journal of Electronics 2024年第3期33卷 584-600页

作者： Xuanhong WANG Cong LI Zengguo SUN Luying HUI Xi'an University of Posts and Telecommunications School of Computer Science Shaanxi Normal University Key Laboratory of Intelligent Computing and Service Technology for Folk Song Ministry of Culture and Tourism

With the rapid development of deep learning,generative adversarial network(GAN) has become a research hotspot in the field of computer *** has a wide range of applications in image *** by GAN,a series of models of Chinese character font generation have been proposed in recent *** this paper,the latest research progress of Chinese character font generation is analyzed and *** and its development history are ***-based methods for Chinese character font generation are clarified as well as their improvements,based on whether the specific elements of Chinese characters are *** public datasets used for font generation are summarized in detail,and various application scenarios of font generation are *** evaluation metrics of font generation are systematically summarized from both qualitative and quantitative *** paper contributes to the in-depth research on Chinese character font generation and has a positive effect on the inheritance and development of Chinese culture with Chinese characters as its carrier.

关键词： Generative adversarial network Font generation image-to-image translation Deep learning

来源：

同方期刊数据库评论

在线全文

同方期刊数据库

学校读者我要写书评

暂无评论

Few-shot font style transfer with multiple style encoders

引用

Science China(Information Sciences) 2022年第6期65卷 83-84页

作者： Kejun ZHANG Rui ZHANG Yonglin WU Yifei LI Yonggen LING Bolin WANG Lingyun SUN Yingming LI State Key Lab of CAD&CG Zhejiang University School of Software Technology Zhejiang University Robotics X Tencent College of Information Science & Electronic Engineering Zhejiang University

Dear editor,From the sketch to an encapsulated font, text font design is a labor-intensive and time-consuming process that relies heavily on the expertise of designers and usually takes months,even years, for a professional institution. This scenario is particularly prominent for glyph-rich scripts, such as Chinese, where each character is composed of varying numbers of highly complex structured components.

关键词： Font Generation image-to-image translation GANs Multi-task Learning Font Fusion Style Transfer

来源：

同方期刊数据库评论

在线全文

同方期刊数据库

学校读者我要写书评

暂无评论

欢迎您,

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

在线全文

在线全文

在线全文

在线全文

在线全文

在线全文

在线全文

请选择保存的检索档案：

请选择收藏分类：

通借通还

欢迎您,

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

在线全文

在线全文

在线全文

在线全文

在线全文

在线全文

在线全文

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：