• 综合
  • 标题
  • 关键词
  • 摘要
  • 学者
  • 期刊-刊名
  • 期刊-ISSN
  • 会议名称
搜索

作者:

Tan, Hongchen (Tan, Hongchen.) | Liu, Xiuping (Liu, Xiuping.) | Yin, Baocai (Yin, Baocai.) (学者:尹宝才) | Li, Xin (Li, Xin.)

收录:

EI Scopus SCIE

摘要:

This article presents a new text-to-image (T2I) generation model, named distribution regularization generative adversarial network (DR-GAN), to generate images from text descriptions from improved distribution learning. In DR-GAN, we introduce two novel modules: a semantic disentangling module (SDM) and a distribution normalization module (DNM). SDM combines the spatial self-attention mechanism (SSAM) and a new semantic disentangling loss (SDL) to help the generator distill key semantic information for the image generation. DNM uses a variational auto-encoder (VAE) to normalize and denoise the image latent distribution, which can help the discriminator better distinguish synthesized images from real images. DNM also adopts a distribution adversarial loss (DAL) to guide the generator to align with normalized real image distributions in the latent space. Extensive experiments on two public datasets demonstrated that our DR-GAN achieved a competitive performance in the T2I task. The code link: https://github.com/Tan-H-C/DR-GAN-Distribution-Regularization-for-Text-to-Image-Generation.

关键词:

text-to-image (T2I) generation Image synthesis Visualization generative adversarial network Generators Training Distribution normalization Stability analysis Semantics semantic disentanglement mechanism Task analysis

作者机构:

  • [ 1 ] [Tan, Hongchen]Beijing Univ Technol, Artificial Intelligence Res Inst, Beijing 100124, Peoples R China
  • [ 2 ] [Yin, Baocai]Beijing Univ Technol, Artificial Intelligence Res Inst, Beijing 100124, Peoples R China
  • [ 3 ] [Liu, Xiuping]Dalian Univ Technol, Sch Math Sci, Dalian 116024, Peoples R China
  • [ 4 ] [Li, Xin]Louisiana State Univ, Sch Elect Engn & Comp Sci, Baton Rouge, LA 70808 USA
  • [ 5 ] [Li, Xin]Louisiana State Univ, Ctr Computat & Technol, Baton Rouge, LA 70808 USA

通讯作者信息:

查看成果更多字段

相关关键词:

来源 :

IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS

ISSN: 2162-237X

年份: 2022

期: 12

卷: 34

页码: 10309-10323

1 0 . 4

JCR@2022

1 0 . 4 0 0

JCR@2022

ESI学科: COMPUTER SCIENCE;

ESI高被引阀值:46

JCR分区:1

中科院分区:1

被引次数:

WoS核心集被引频次: 21

SCOPUS被引频次: 28

ESI高被引论文在榜: 0 展开所有

万方被引频次:

中文被引频次:

近30日浏览量: 4

归属院系:

在线人数/总访问数:505/4931912
地址:北京工业大学图书馆(北京市朝阳区平乐园100号 邮编:100124) 联系我们:010-67392185
版权所有:北京工业大学图书馆 站点建设与维护:北京爱琴海乐之技术有限公司