Deep Learning Fundamentals - Premium Edition

Image Data As Neural Network Input Explained - Deep Learning

We've talked a lot about passing images to neural networks as input, and we've also learned all about how to pre-process image data before we pass them to networks. We'll now explore how exactly image data is received as input by the network, and how it transforms as it traverses through the network layers.

During the data pre-processing step that occurs before training, we know that image data is saved on disk that will later be passed to the network as input. When we actually pass image data to a network as input, however, we don't just pass the saved image files. Instead, we pass the data broken down by its features.