A nice programming challenge is writing a program that modifies an image to give using the style from another image, which in my case is Van Gogh’s Starry Night.
Let’s take a look at the input image.
A take of the houses of parliament at night.
Now let’s see the output image after 7000 iterations.
How is it possible? The AI Artist gathers information from the VGG19, a very deep image recognition convolutional neural network with 19 layers, to mix the content of one image with the style of another.
In such problems, it’s important to find an adequate loss function and to minimize it. To do it, I used the standard Adam Optimizer. Each iteration, or learning step, is meant to reduce such a loss function, which is the ideal distance from a perfect blend of the two images.
You can find more information and the code on:
https://github.com/mattiagiuri/machine-learning-projects/tree/main/Image%20Modifier