Let’s say, we want to build an object detection classifier.
First, we need to prepare a training set
Build a conv net and train it to detect if there is an object (car in this case)
Repeat woth a larger window throughout the image.
Repeat with even larger window.