Summary
CNNs are the go-to model for capturing patterns in image data. The handpicked architectures that were introduced in this chapter are the core backbones that can be subsequently utilized as a base for solving more custom downstream tasks such as image object detection and image generation.
The CNNs that were covered here will be used practically in later chapters as a basis to help you learn other deep learning-based knowledge. Take your time and look into how different architectures are implemented in a deep learning library offline in this book’s GitHub repository; we won’t be presenting the actual implementation code here. Now that we have covered CNNs in intermediate to low-level detail, in the next chapter, we’ll shift gears and look at recurrent neural networks.
