In this article, I want to talk about the use of convolutional neural networks for the classification of images by style.
The goal of our project is to build software to identify whether an image is in the “BMW style”. In other words, we are faced with the task of classifying images. It is important to note here that images could be of any content, with and without cars. So, the main interest here is not to identify a car object, or identify a BMW car, rather identify a BMW look and feel – colors, composition and so on. But we can’t select these attributes of style manually. To solve this problem, it was proposed to use a neural network, in which such complex features will be found automatically in the learning process.