Technical requirements

Google Colab notebooks: Because we will be training a convolutional NN (CNN) on slightly larger datasets than our laptop CPU could handle, we will develop our models in Google Colab. Colab is a free-to-use Jupyter notebook-kind of framework that provides a free Graphical Processing Unit (GPU) to train models. We will learn the basics of Google Colab in the following section, but readers can find more details about Colab here: https://towardsdatascience.com/getting-started-with-google-colab-f2fff97f594c. The key to building models in Google Colab is selecting a GPU so that you can build your model faster. You just go to the Runtime drop-down menu, select Change Runtime type, and then select GPU in the Hardware accelerator drop-down menu, as shown here:

Figure 9.1 – Google Colab notebook settings

Once a GPU has been selected, you can run the commands shown in the hands-on section of this chapter.