Exploring the CNN architecture families

Now, instead of going through the history of CNN through the years, let’s look at a list of the different handpicked model architecture families. These architecture families are selectively chosen to be sufficiently different and diverse from each other. One thing to note is that neural networks are advancing at an astounding pace. With that in mind, the architecture families that will be introduced here are ensured to be relevant today. Additionally, only the most important information you need to know about the architecture family will be presented, simplifying the many pages of research papers in concise but sufficient detail.

Another thing to note before diving into this topic is that the metric performance on a dataset will often be the main comparison method among different architectures, so be aware that the metric performance of a model is achieved by the collective contribution of the training method and the architecture. The...