The KNN algorithm is very susceptible to local noise and isn't very useful when there is a lot of overlap between classes expected. It is typically not very useful for more advanced tasks, such as psychographic, demographic, or behavioral analysis. But it's a very useful tool to keep handy in your toolbox, because it can assist with lower-level tasks very easily.
In this example, we'll use our KNN class to de colorize a photo. Specifically, we're going to take colorful input photos and restrict them to a color scheme of only 16 colors. We'll use KNN here to select the appropriate replacement color for a pixel, given that pixel's original color.
Our workflow will look like this:
-
Use the jimp library to read ...
