MachineBox is a machine learning system as a service. It has, presumably, in some backend somewhere, a general model—say, a convolutional neural network that has been trained with many faces, such that it knows what a face is. It does not provide the specific model that you may require for the task at hand. So instead, we would need to fine-tune the model provided by MachineBox by giving it training data. Per MachineBox's terminology, this is called teaching. As part of a curiosity collection, I have collected a small but usable number of images of MC Hot Dog's face that are suitable for the task of teaching the MachineBox what MC Hot Dog looks like.
For this project, the images are in the hotdog.zip file. Unzip the file into a ...
