Conv layers filters doubt


in the first slides explaining CNN, we convoluted pictures with specific filters and depending on the filter we were detecting vertical edges, horizontal edges,…
Later in the course when we start to use more filters and the filters are volumes (usually depth 3 because of RGB) we stop talking about the filters cell values. We add layers and layers of several filters.

What’s the content of those filters? Is it a default set of filters from Tensorflow for each filter size f? Am I missing something?

Thank you

The contents of the filters is what is learned during training.

TensorFlow hides all of the details of the training process behind the scenes.

You are absolutely right. Got it. There are so many concepts that some times I need to make one step back.

Thank you.