OCR comparison DevonTHINK vs ABBY FineReader

Are there any comparison studies of OCR accuracy?

I have a Fujitsu ScanSnap and it came with ABBy FineReader. Should I use the ABBY FineReader OCR, or the OCR engine built into DevonTHINK Pro? Which is more accurate?

Anyone? Bill?

As I haven’t found any comparative reviews and don’t have FineReader for OS X, I can’t respond beyond saying that.

While we have no comparison of OCR engines, we have at least considered evaluating the Abbyy engine for DEVONhink Pro Office as its easier and more seamless to embed into the application. However, the Abbyy engine is only available for Intel processors and quite a few of our users still use PowerPC based Macs so we’ll have to wait a bit longer.

Thanks for the response guys…

I have another question…
I scan my paper about once a week. Running the OCR while scanning isn’t an option. It’s just too slow. So I use my ScanSnap to make non-OCRed pdfs. I name the PDFs. I bring them into various folders in my DevonTHINK records archive.

Then, I search for “pdf” and organize by type. I select all of the “pdf” files, but don’t select the “pdf+text” files. Then I run the OCR in DevonTHINK using the “convert to searchable pdf” function on them and letting it run overnight.

I have 4 questions:

1– Right now, once the OCR in DTP is finished, I have both “pdf” and “pdf+text” versions in my database. Is there any way to (easily) get DevonTHINK to erase the original pdf automatically once the “pdf+text” is created?

2– Would I get better results using the ABBY Finereader before importing into DevonTHINK? (I realize you probably can’t answer this any more than you already have.)

3– What ScanSnap settings should I use for the initial scan in order to achive optimal OCR results in DevonTHINK OCR later?

Also - in the help file I read “Note: To save disk space, the generated searchable PDF is rendered in 150 dpi.”… Does this apply ONLY when scanning DIRECTLY into DTP? Or does it also apply when I use the “convert to searchable PDF” option? If I scan outside of DTP at 600 DPI, then import and run the “convert to searchable PDF”, will the “pdf+text” rendered be 600 DPI or 150 DPI? (I would want it to be 150 DPI, my only reason for scanning higher would be to get better OCR results.

Answers to some of your questions:

  1. Why not disable “Set Attributes” in the OCR preference pane and the documents will be processed in the background without a panel coming up? Otherwise, your only solution is to go to AppleScript to automate this.

  2. Colour, 300 dpi for “standard” typeface documents.

The image dpi & quality settings apply to all OCR result documents.