We learned the classical feature-based approach to handling images in a machine learning context; by converting from a million pixels to a few numeric features, we were able to directly use a logistic regression classifier. All of the technologies that we learned in the other chapters suddenly became directly applicable to image problems. We saw one example in the use of image features to find similar images in a dataset.

We also learned how to use local features in a bag of words model for classification. This is a very modern approach to computer vision and achieves good results, while being robust enough for many irrelevant aspects of the image, such as illumination, and even uneven illumination in the same image. We also used clustering as a useful intermediate step in classification, rather than as an end in itself.

We focused on mahotas, which is one of the major computer vision libraries in Python. There are others that are equally well maintained. Skimage is similar in spirit, but has a different set of features. OpenCV is a very good C++ library with a Python interface. All of these can work with NumPy arrays and you can mix and match functions from different libraries to build complex computer vision pipelines.

We also tried a new way of generating similar images with Tensorflow (which can be used on non-image domains) with the current trendy type of network named GAN.

In Chapter 13, Reinforcement Learning, we will explore reinforcement learning, a hot topic for deep learning. We will see how we can make a neural network learn a set of rules without telling it anything.