I'm using the prebuilt model to extract invoice data.
I shifted from v2 to v3:
1) Vendorname recognition/parsing is better
2) TotalTax gets missed more ...
Anyway to circumvent this?
I'm using the prebuilt model to extract invoice data.
I shifted from v2 to v3:
1) Vendorname recognition/parsing is better
2) TotalTax gets missed more ...
Anyway to circumvent this?
@philipverlinden Is it possible to post the image used? I think this is most likely because OCR did not run as expected on the image you are using.
Yes, I can provide you a copy of the PDF.
How can I do this on a secured way please?
Thank you!
Philip
@romungi-MSFT forgot to tag/mention you in the previous reply.
Yes, I can provide you a copy of the PDF.
How can I do this on a secured way please?
Thank you!
Philip
9 people are following this question.