i have used invoice ml package and trained 20 document for invoice no and created ml skills .Now when using ml skill from studio using document understanding and trying to extract value of invoice for the document which I have trained .it is not able to extract proper invoice no.Can i know how to ensure better accuracy of field .
Not extracting proper invoice value from the document after giving training on 20 document using Datamanager
Usually in document understanding we can use three different ways of getting a text and all ways can be enabled so that the one with high accuracy will be considered as final output
So in that case apart from ML we can try using REGEX, FORM EXTRACTOR AND INTELLIGENT FORM EXTRACTOR combined together
Thank you for your answer.But I am using Datamanager to train the field which is not available in inbuilt document understanding so in that case if we train different types of document and after training from datamanger if we try to extract same document value if it not getting extracted then what is the use of training that all i need to know if document is diffrent for whcih require training then in that case how we can use regular expression.because the value extracted showing confidence level greater than 90% then also it is incorrent.Then how we will ensure data is accurate or not
Hi there, I invite you to try Nanonets Cognitive OCR (Our UiPath connector: NanoNets OCR - RPA Component | UiPath Marketplace) We use AI specifically to handle various document variants. You can directly process invoices without writing templates/rules.