Generate ground truth cases for object localization algorithms.
Cycle through a list of images, select points around which to generate bounding
boxes and assign classifiers. Output the coordinates, and images annotated with
boxes and labels. For an example study that uses bounding boxes for image
localization and classification see Ibrahim, Badr, Abdallah, and Eissa (2012)
"Bounding Box Object Localization Based on Image Superpixelization"