We apply convolutional neural networks (CNN) to the problem of image
orientation detection in the context of determining the correct orientation
(from 0, 90, 180, and 270 degrees) of a consumer photo. The problem is
especially important for digitazing analog photographs. We substantial