So far, we have covered the creation of dataset objects and how to create batches of data to feed into a model. In this section, we will look at an example of an end-to-end input data pipeline and model training. We will build an image classifier using the CIFAR10 data.
In order to run the CIFAR10-based end-to-end example, you need to download the necessary data from https://www.cs.toronto.edu/~kriz/cifar.html. The dataset has been taken from a paper called Learning Multiple Layers of Features from Tiny Images, by Alex Krizhevsky, 2009 (https://www.cs.toronto.edu/~kriz/learning-features-2009-TR.pdf). This dataset contains the following information:
- 50,000 images with labels for training
- 10,000 images with labels for testing
- 10 class labels
After downloading and untarring the dataset, you will see a folder called cifar-10-batches...
