R Deep Learning Essentials - Second Edition

Image classification models

We covered image classification in Chapter 5, Image Classification Using Convolutional Neural Networks. In that chapter, we described convolutional and pooling layers that are essential for deep learning tasks involving images. We also built a number of models on a simple dataset, the MNIST dataset. Here, we are going to look at some advanced topics in image classification. First, we will build a complete image classification model using image files as input. We will look at callbacks, which are a great aid in building complex deep learning models. A call-back function will be used to persist (save) a model to file, which will be loaded back later. We then use this model in our next example, which is transfer learning. ...