Convolutional Neural Networks

In this lesson we are going to take a deeper dive in to CNNs and their variants. Almost all imagery projects will use CNNs in one form or another.

What is a convolution

Remeber from last time, CNNs have their weights arranged into blocks called filters. This forces the weights to have an explicit spatial relationship.

One layer of a CNN will be composed of many filters (sometimes called kernels). Each filter produces one output.