June 25, 2015
This experiment demonstrates the usage of a two-class neural network to classify images.
This experiment is designed with a TSA-like scenario in mind. The dataset contains images of X-rayed luggage, some with dangerous objects, some without. A \"manual\" classification is provided for training purposes, then a two-class neural network model and a two-class perceptron are trained and tested. \nThe image reader block is limited in what it can do, so the images must be all the same size and relatively small (256 x 256 or 65536 pixels total). The pictures are in a public, read-only blob container.