Training

Training is the process of changing the OCR solutions assigned to character shapes in the image. It serves to improve OCR accuracy on long and similar styled documents. It could be worth training any repeatedly misrecognized, unrecognized or uncommon character or symbol. Automatic training, called IntelliTrain, derives training data from the corrections a user makes while proofing text from the Text Editor. Manual training allows user-selected characters to be trained after recognition.

All training can be saved to a training file for future use and loaded whenever needed. A training file contains a set of character shapes each associated with an OCR solution. When a training file is loaded, these stored shapes are compared with problem shapes found on pages being recognized. The assigned solutions will be applied when appropriate.

Training data can exist and be used without it being saved to a training file. Unsaved training data is conserved inside an OmniPage Document. If you do not save your document as an OPD file and do want the training data in a file, you must save it before you close the document. Whenever you close a document or switch to another one when unsaved training data exists, a dialog box appears allowing you to save it.

Training cannot be applied to Japanese, Korean or Chinese. Do not do manual training, and before recognizing these languages, ensure than IntelliTrain is disabled in the Proofing panel of the Options dialog box. Check also that no training file is loaded, since Asian OCR would risk compromising its contents.

//

Training