A Convolutional Neural Network (CNN) is a self-learning network that classifies images similar to how our human brain learns, by observing images of different classes. CNNs learn the content of an image by applying image filtering and by processing the methods of various filter size, quantity, and non-linear operations. These filters and operations are applied across many layers so that the spatial dimensions of each subsequent layer decrease and their depths increase during the image transformation process.
For each filtering application, the depth of the content that's learned increases. This starts with edge detection, followed by recognizing shapes, and then a collection of shapes called features, and so on. This is analogous to the human brain when it comes to how we comprehend information. For example, during a test on reading...