OCR Engines performance

Help Document Understanding
,
1

Hi,

As of now, the following are the OCR engines available in UI Path.

1.Microsoft
2.Omnipage
3.Tesseract
4.Abby
5.Abby Cloud
6.Microsoft Project Oxford Online
7.Microsoft Azure Computer Vision
8.Google Cloud Vision

The question is which engine is best suitable to read Scanned (image based) PDFs as of now?

I have found this article and tested few samples of scanned documents with all above mentioned OCRs except Abby.

Can you en-light based on your past work on OCRs ,which will help me in choosing the right engine (Free on-premises or Paid on-premises or Cloud)?

2

Hello @Velthurai,

There is one more, the OmniPage OCR activity found in the UiPath.OmniPage.Activities package.

Unfortunately I cannot provide a comparison between the engines, as each of them has their advantages and disadvantages, and the best approach for choosing one is to actually gather a relevant sample set for your specific use case, and test them all out. This is the safest and sanest approach to go with, as in some cases one engine might perform better than others.

Ioana

1 Like