In this chapter, we explored image data and a deep neural network image-classification model. We used data from 18 images of bicycles, cars, and airplanes, and carried out appropriate data processing to make the data ready to use with the Keras library. We partitioned image data into training, validation, and test data, and subsequently developed a deep neural model using training data and evaluated its performance by looking at the loss, accuracy, confusion matrix, and probability values for both the training and test data. We also made modifications to the model to improve its classification performance. In addition, we observed that when the confusion matrix provides the same level of performance, prediction probabilities may be able to help in extracting finer differences between the two models.

In the next chapter, we will go over the steps to develop a deep neural network image-classification model using convolutional neural networks (CNNs), which are becoming very popular when it comes to image classification applications. CNNs are regarded as the gold standard for image-classification problems, and are very effective for large-scale image-classification applications.