15 December 2017
Sorry, probably there was some misunderstanding in stages' numeration. The workflow schema should look like
- Recognition1 Stage (or default recognition). FL1 and FL2 are matched and recognized, all other documents become <Unknown>
- Script Stage (with entry condition that allows only <Unknown> documents inside) - cleans recognition results by IBatch::CleanRecognitionResults( document : IDocument ) method, then sets IMatchingInfo:efinitionsList to FL3 and ForceMatch to true
- Recognition2 Stage (linked to previous, so only ex-<Unknown> documents are admitted) forcematches the FL3 layout. If no OCR is really needed, then IMatchingInfo::NeedRecognition could be set up to false on the previous script stage, that will allow to speed up the processing.
So we have 3 stages, but only two of them are Recognition, and the Recognition2 is needed to create necessary fields in the documents.