> ## Content Index
> Fetch the complete content index at: https://sudormrf.run/llms.txt
> Use this file to discover other available public pages before exploring further.

# Image Translation with GAN (1)
- URL: https://sudormrf.run/2018/04/01/image-translation-with-gan-1/
- Published: 2018-04-01T08:05:56.000Z
- Updated: 2018-04-05T12:20:14.000Z
- Author: Junho Cho
- Tags: Computer Vision, Deep Learning, #Import 2026-10-24 03:36

# Problem statement of Image Translation

![inline](https://tmmsexy.s3.amazonaws.com/imgs/2017-05-10-120608.jpg)

### Learn $G: (S \\rightarrow T)$

$G$ that convert an image of source domain $S$ to an image of target domain $T$  
Domain Adaptation/Transfer of an image.

### Paired and Unpaired dataset

![inline](https://ws2.sinaimg.cn/large/006tNc79gy1fpa9gc6l7bj30ds08ut9i.jpg)

### Image Translation: $S$ and $T$ are pair-wise labeled in dataset

![inline](https://tmmsexy.s3.amazonaws.com/imgs/2017-05-10-120245.jpg)

### Image Translation: $S$ and $T$ are not pair-wised in dataset

![inline](https://tmmsexy.s3.amazonaws.com/2017-03-30-121602.jpg)

![fit](https://github.com/junyanz/CycleGAN/raw/master/imgs/horse2zebra.gif)  
![inline](https://junyanz.github.io/CycleGAN/images/objects.jpg)

# Before, Style Transfer (NeuralArt) was prominent

![](https://tmmsexy.s3.amazonaws.com/imgs/2017-05-10-120857.jpg)

But it largely depends on textual information of an target style

![inline fit](https://tmmsexy.s3.amazonaws.com/imgs/2017-05-10-121911.jpg)  
![inline fit](https://tmmsexy.s3.amazonaws.com/imgs/2017-05-10-121932.jpg)

# How to learn more general Image Translation?

### Generative Adversarial Network

![fit right](https://raw.githubusercontent.com/torch/torch.github.io/master/blog/_posts/images/out.gif)  
![inline](https://cdn-images-1.medium.com/max/800/1*-gFsbymY9oJUQJ-A3GTfeg.png)  
![inline](https://tmmsexy.s3.amazonaws.com/2017-03-30-070136.jpg)

\\begin{equation}  
\\min\_G \\max\_D V(D,G) = \\mathbf{E}\_{x\\sim p\_{\\text{data}}(x)} \[\\log D(x)\] + \\mathbf{E}\_{z\\sim p\_{z}}(z)\[\\log (1 - D(G(z)))\].  
\\end{equation}

### also honorable mention : Deep Convolutional GAN (DCGAN)

# Two major problems of Image Translation

1. Convert to which domain?  
  - learn which "$G: (S \\rightarrow T)$"?
2. How to learn the dataset?  
  - how to properly form dataset?
  - pair-wise Supervised? or Unsupervised?

### Today, presenting SOTA of Image Translation papers of

- [pix2pix](https://sudormrf.run/2018/04/02/image-translation-with-gan-2/): Image-to-Image Translation with Conditional Adversarial Networks (CVPR2017)
- [Domain Transfer Network](https://sudormrf.run/2018/04/02/image-translation-with-gan-3/): Unsupervised Cross-Domain Image Generation (ICLR2017)
- [CycleGAN & DiscoGAN](https://sudormrf.run/2018/04/03/image-translation-with-gan-4/): CycleGAN: Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks (ICCV2017) & DiscoGAN: Learning to Discover Cross-Domain Relations with Generative Adversarial Networks (ICML2017)
- [BicycleGAN](https://sudormrf.run/2018/04/04/image-translation-with-gan-5/): Toward Multimodal Image-to-Image Translation (NIPS2017)