文章主要整理了GAN网络及其各种变体模型,并给出了模型的论文出处及代码实现,结合最原始的论文和代码实现,可以加深对模型原理的理解。
目录
GAN
Auxiliary Classifier GAN
Bidirectional GAN
Boundary-Seeking GAN
Context-Conditional GAN
Coupled GANs
CycleGAN
Deep Convolutional GAN
DualGAN
Generative Adversarial Network
InfoGAN
LSGAN
Semi-Supervised GAN
Wasserstein GAN
GAN
实现最原始的,基于多层感知器构成的生成器和判别器,组成的生成对抗网络模型(Generative Adversarial)。
参考论文:《Generative Adversarial Networks》
代码地址:https://github.com/eriklindernoren/Keras-GAN/blob/master/gan/gan.py
AC-GAN
实现辅助分类-生成对抗网络(Auxiliary Classifier Generative Adversarial Network)。
参考论文:《Conditional Image Synthesis With Auxiliary Classifier GANs》
代码地址:https://github.com/eriklindernoren/Keras-GAN/blob/master/acgan/acgan.py
BiGAN
实现双向生成对抗网络(Bidirectional Generative Adversarial Network)。
参考论文:《Adversarial Feature Learning》
代码地址:https://github.com/eriklindernoren/Keras-GAN/blob/master/bigan/bigan.py
BGAN
实现边界搜索生成对抗网络(Boundary-Seeking Generative Adversarial Networks)。
参考论文:《Boundary-Seeking Generative Adversarial Networks》
代码地址:https://github.com/eriklindernoren/Keras-GAN/blob/master/bgan/bgan.py
CC-GAN
实现基于上下文的半监督生成对抗网络(Semi-Supervised Learning with Context-Conditional Generative Adversarial Networks)。
参考论文:《Semi-Supervised Learning with Context-Conditional Generative Adversarial Networks》
代码地址:https://github.com/eriklindernoren/Keras-GAN/blob/master/ccgan/ccgan.py
CoGAN
实现耦合生成对抗网络(Coupled generative adversarial networks)。
参考论文:《Coupled Generative Adversarial Networks》
代码地址:https://github.com/eriklindernoren/Keras-GAN/blob/master/cogan/cogan.py
CycleGAN
实现基于循环一致性对抗网络(Cycle-Consistent Adversarial Networks)的不成对的Image-to-Image 翻译。
参考论文:《Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks》
代码地址:https://github.com/eriklindernoren/Keras-GAN/blob/master/cyclegan/cyclegan.py
DCGAN
实现深度卷积生成对抗网络(Deep Convolutional Generative Adversarial Network)。
参考论文:《Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks》
代码地址:https://github.com/eriklindernoren/Keras-GAN/blob/master/dcgan/dcgan.py
DualGAN
实现对偶生成对抗网络(DualGAN),基于无监督的对偶学习进行Image-to-Image翻译。
参考论文:《DualGAN: Unsupervised Dual Learning for Image-to-Image Translation》
代码地址:https://github.com/eriklindernoren/Keras-GAN/blob/master/dualgan/dualgan.py
InfoGAN
实现的信息最大化的生成对抗网络(InfoGAN),基于信息最大化生成对抗网络的可解释表示学习。
参考论文:《InfoGAN: Interpretable Representation Learning by Information Maximizing Generative Adversarial Nets》
代码地址:https://github.com/eriklindernoren/Keras-GAN/blob/master/infogan/infogan.py
LSGAN
实现最小均方误差的生成对抗网络(Least Squares Generative Adversarial Networks)。
参考论文:《Least Squares Generative Adversarial Networks》
代码地址:https://github.com/eriklindernoren/Keras-GAN/blob/master/lsgan/lsgan.py
SGAN
实现半监督生成对抗网络(Semi-Supervised Generative Adversarial Network)。
参考论文:《Semi-Supervised Learning with Generative Adversarial Networks》
代码地址:https://github.com/eriklindernoren/Keras-GAN/blob/master/sgan/sgan.py
WGAN
实现 Wasserstein GAN。
参考论文:《Wasserstein GAN》
代码地址:https://github.com/eriklindernoren/Keras-GAN/blob/master/wgan/wgan.py
领取专属 10元无门槛券
私享最新 技术干货