This uses the k-means algorithm to find the closest colors and group them based on the specified number of colors. It also uses the naive sharding centroid initialization method to initalize the centroids.