iPython Notebooks

Running OCRopus

Ocropus Steps illustrates the basic steps in running OCRopus, as well as the intermediate representations and how to access them from Python.

Making Ground Truth

Making Groundtruth shows how to generate ground truth for training from large amounts of unlabeled training data.

Normalization

Normalization is an important preprocessing step for HMM and RNN recognizers. There is a particular set of classes and APIs in OCRopus to perform this normalization. They are illustrated in these notebooks: Simple Normalization and Normalization API