8

I'm just getting started with these topics. To the best of my knowledge, style transfer takes the content from one image and the style from another, to generate or recreate the first in the style of the second whereas GAN generates completely new images based on a training set.

But I see a lot of places where the two has been used interchangeably, like this blog here and other places where GAN is used to achieve style transfer, like this paper here

Are GAN and Style transfer two different things or is GAN the method to implement style transfer or are they both different things that does the same thing? Where exactly is the line between the two?

4

2 回答 2

5

GAN is a neural network architecture

style transfer is a (set of) processing method (can be as simple as grayscale or blur)

So the relation is:

  • GAN can be used to implement style transfer. (and other things)

To make it more complicate (hopefully this can make something clear), if you think the feature vector as a style of an image, then feature vector -> image conversion is a style transfer :)

于 2019-03-11T15:28:26.780 回答
1

GAN is more generalized model than Style Transfer. Both the methods try to solve the same problem but the approach is different. Style transfer tries to keep the content of the image intact while applying the style of the other image. It extracts the content and style from middle layers of the NN model. It focusses on learning the content and style of the image separately but in GAN, the model tries to learn the entire mapping from one domain to another without segregating the learning of context and style.

loss function of style transfer = Loss(content) + Loss(style)
loss function of GAN = loss(cycle consistency) + loss(adversarial)

于 2020-07-24T19:02:33.907 回答