5

Advanced Computer Vision Applications

In Chapter 4, we introduced convolutional networks (CNNs) for computer vision and some of the most popular and best-performing CNN models. In this chapter, we’ll continue with more of the same, but at a more advanced level. Our modus operandi so far has been to provide simple classification examples to support your theoretical knowledge of neural networks (NNs). In the universe of computer vision tasks, classification is fairly straightforward as it assigns a single label to an image. This also makes it possible to manually create large, labeled training datasets. In this chapter, we’ll introduce transfer learning (TL), a technique that will allow us to transfer the knowledge of pre-trained NNs to a new and unrelated task. We’ll also see how TL makes it possible to solve two interesting computer vision tasks – object detection and semantic segmentation. We can say that these tasks are more complex compared to classification...