Style Transfer

In simple words, style transfer consists of modifying the style of an image while still preserving its content. One example would be taking an image of an animal and transforming the style into a Monet-like painting, as shown in the following image:

Figure 5.1: Style transfer inputs and output – the result of the final exercise of this chapter

Note

This image is available on GitHub at https://packt.live/2XEykpL.

According to the preceding image, there are two inputs to the model: a content image and a style image. Content refers to the objects of an image, while style refers to the colors and textures. As a result, the output from the model should be an image containing the objects from the content image and the artistic appearance of the style image.

How Does It Work?

As opposed to solving a traditional computer vision problem, which was explained in the previous chapter, style transfer requires a different set of steps...