During the last few years, several apps have been appearing that make a caricature from a sent photograph, or that detect the outline of an image to make a drawn version of it. Many of these apps no longer exist (it is difficult to find a solid business model in this category), but new options keep appearing.
Today we talk about a different one, since the result is neither realistic, nor credible, nor of good quality, but it is created by an Artificial Intelligence, which helps us to better understand how some techniques work.
Cartoonify uses a neural network to turn the loaded photo into a unique cartoon. In the same way that the Draw This camera converted the images it captured into cartoons thanks to a Raspberry Pi, this site allows anyone to try to caricature any image, and for that it uses the drawings that Google saves in its database. In the drawing web of Google we can see how we recognize what we draw, so that the flow of information from Cartoonify is simple: it identifies what is in our photo and verifies what people draw when they ask for exactly that.
It is recommended to use images with clear objects on a solid background, although the result is usually equally disastrous.