In this chapter, we discussed some popular CNN architectures: we started with the classics, AlexNet and VGG. Then, we paid special attention to ResNets, as one of the most well-known network architectures. We also discussed the various incarnations of Inception networks and the Xception and MobileNetV2 models, which are related to them. We also talked about the exciting new ML area of neural architecture search. Finally, we discussed capsule networks—a new type of CV network, which tries to overcome some of the inherent CNN limitations.
We've already seen how to apply these models in Chapter 2, Understanding Convolutional Networks, where we employed ResNet and MobileNet in a transfer learning scenario for a classification task. In the next chapter, we'll see how to apply some of them to more complex tasks such as object detection and image segmentation...
