Decipher IDP OCR Training

[MobeenAli]

Hi, I wanted to know does Decipher perform training for the OCR or does it only train the Models which Capture the fields from which data is to be extracted?

From what I understood until now is that Decipher only trains the capture model which if Machine Learning is turned on trains the ML model or in case Machine Learning is off trains the rule-based model.

[Rambo27]

Hi Mobeen Ali, 

For Decipher -

https://bpdocs.blueprism.com/decipher-1-2/en-us/user-guide/user-guide.htm

All relevant things available to start with .

[Ben.Lyons1 · Accepted solution]

Hi Mobeen,

Decipher doesn't "train" the OCR as such, it uses an OCR engine to extract characters and words from the document.

It then creates what's called "Training Data", which is based on what it has learned from the DFD configuration in relation to what has been identified in the document. This is created via the user input during the data verification stage. It is not directly linked to the DFD/Document Type and is always on

The same process as above can be used to inform an ML Capture Model. This is specific to a Document Type and is enabled on user request. Please  visit our Best Practice guide for more information.

Thanks

Ben