Where can I get an explanation of convolutional neural networks using 3D matrices as an example?

Question

Neural networks

Probably a funny question, but I can’t find an explanation of the operations used in CNN using the example of three-dimensional matrices. The Internet is full of super-obvious pictures explaining convolution and aggregation using simple single-band images as an example, but there is no explanation of what to do after convolution with several filters.

Let's say we have a grayscale image with a resolution of 28x28 . We perform convolution with 8 3x3 filters and get a 26x26x8 matrix . And then what? How to do further convolution or union on such a matrix? What article / literature (preferably in Russian) do you recommend reading for a full understanding of how CNN works ?

Answer 1

You have 8 layers in the convolution core for gray, add 8 more for red and 8 more for green.
No, I lied. The number of layers in the convolution is equal to the dimension of the input. And the same light processes all incoming channels.