esrgan_ESRGAN: Enhanced Super-Resolution Generative Adversarial Networks

梦里梦外; 2023-01-10 05:12 119阅读 0赞

摘要：

The Super-Resolution Generative Adversarial Network (SRGAN) is a seminal work that is capable of generating realistic textures during single image super-resolution. However, the hallucinated details are often accompanied with unpleasant artifacts. To further enhance the visual quality, we thoroughly study three key components of SRGAN - network architecture, adversarial loss and perceptual loss, and improve each of them to derive an Enhanced SRGAN (ESRGAN). In particular, we introduce the Residual-in-Residual Dense Block (RRDB) without batch normalization as the basic network building unit. Moreover, we borrow the idea from relativistic GAN to let the discriminator predict relative realness instead of the absolute value. Finally, we improve the perceptual loss by using the features before activation, which could provide stronger supervision for brightness consistency and texture recovery. Benefiting from these improvements, the proposed ESRGAN achieves consistently better visual quality with more realistic and natural textures than SRGAN and won the first place in the PIRM2018-SR Challenge. The code is available at this https URL .

展开

发表评论取消回复

表情：

评论列表（有 0 条评论，119人围观）

还没有评论，来说两句吧...

相关阅读

相关生成对抗网络中的样本生成（Sample Generation in Generative Adversarial Networks）

![275282467ba7408abfb3fa6acdacdcf9.png][] 目录深度学习算法中的生成对抗网络中的样本生成（Sample Generation in

比眉伴天荒/ 2024年02月29日 04:07/ 0 赞/ 52 阅读

相关生成对抗网络GAN：Generative Adversarial Networks

[原文链接][Link 1] 1 原理对于生成对抗网络GAN，一个简单的理解是可以将其看做博弈的过程，我们可以将生成模型和判别模型看作博弈的双方，比如在犯罪分子造假币

浅浅的花香味﹌/ 2023年10月08日 09:01/ 0 赞/ 64 阅读

相关 GAN：生成对抗网络 Generative Adversarial Networks

[原文链接][Link 1] 1 原理对于生成对抗网络GAN，一个简单的理解是可以将其看做博弈的过程，我们可以将生成模型和判别模型看作博弈的双方，比如在犯罪分子造假币

桃扇骨/ 2023年09月28日 16:52/ 0 赞/ 161 阅读

相关 Triple Generative Adversarial Nets

![watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3lxbF82

本是古典何须时尚/ 2023年07月22日 05:58/ 0 赞/ 7 阅读

相关 esrgan_ESRGAN: Enhanced Super-Resolution Generative Adversarial Networks

摘要： The Super-Resolution Generative Adversarial Network (SRGAN) is a seminal work that

梦里梦外;/ 2023年01月10日 05:12/ 0 赞/ 120 阅读

相关【深度学习】Generative Adversarial Networks ,GAN生成对抗网络分类

目录 1 思维导图 2 大纲 3 经典GAN简介 3.1 SGAN 3.2 Conditional GAN 3.

小咪咪/ 2022年10月05日 01:57/ 0 赞/ 405 阅读

相关 Semi and Weakly Supervised Semantic Segmentation Using Generative Adversarial Network

Semi and Weakly Supervised Semantic Segmentation Using Generative Adversarial Network 3

港控/mmm°/ 2022年04月10日 04:29/ 0 赞/ 182 阅读

相关 Conditional Generative Adversarial Network for Structured Domain Adaptation

Conditional Generative Adversarial Network for Structured Domain Adaptation Abstract 近

谁践踏了优雅/ 2022年04月02日 03:44/ 0 赞/ 172 阅读

相关 SentiGAN: Generating Sentimental Texts via Mixture Adversarial Networks论文笔记

另一篇很好的解释：[https://www.itcodemonkey.com/article/6378.html][https_www.itcodemonkey.com_art

桃扇骨/ 2022年03月18日 02:12/ 0 赞/ 192 阅读

相关《SentiGAN: Generating Sentimental Texts via Mixture Adversarial Networks》论文笔记

自己的总结：这篇文章主要是写了一个用多分类器和多个生成器生成带情感的文本的gan模型，结果最后的实验不是生成器的个数还是设为2，相当于并没有比之前的工作多了类型，另外和

系统管理员/ 2022年01月12日 19:39/ 0 赞/ 255 阅读