How Does The OCR SDK Compare To Competitor Products?

[aitchisj]

I'm at a make-it or break-it point in time with this OCR SDK. I'm able to make it work and I want nothing less than for this product to be the one that we choose. My boss asked me how long it takes on average to OCR a page with this control so I performed a few tests to get a ballpark figure based on some test data that I've been working with. For the files that I was testing with, I was getting results in the range of about 4-8 seconds per page, depending on the document. I'm scanning using 300 DPI, no whitelist, no blacklist, and allow for auto rotation. I'm not using zonal scanning either, I'm scanning the entire document. Perhaps you can suggest some tips to improve the performance I'm getting without sacrificing the OCR quality?

I don't know if that's typical or not (4-8 seconds per page), but my boss did suggest it needs to be much less than 5 seconds per page, like 2.5 would be more like what he's looking for. I realize that a number of factors play into how long it takes to OCR, things like hardware and the document itself... so this testing result isn't really the holy grail answer, but it does give me some clues as to what I can expect. I'm playing around with some of the competitor products out there to compare my results; however, I thought I'd ask you guys as well. Surely you do much more comprehensive testing than I can do and can answer some of these questions better than I can.

In terms of speed, how does the OCR SDK compare to competitor products?
In terms of accuracy, how does the OCR SDK compare to competitor products?

To me, it's not a deal breaker if your OCR SDK can only perform at 80% of some of the competitor products -- I realize this is a work in progress and things will improve; however, anything less than 80% might require me to consider alternatives.

Please provide me with any information you can about this as it will be helpful for me and my organization to make the right decision. Thanks in advance,

-John

[Walter-Tracker Supp]

On the whole our OCR compares favourably with our competitors...

On the standard test documents we use, we sit pretty much in the dead middle for speed (slightly faster than our big name PDF competitor, slightly slower than the top of the line OCR-only suite ). Speed depends quite a bit on input document complexity, of course.

We have some speed advances coming up in the next version of the viewer (and SDK).

Accuracy-wise, we are certainly competitive. As with all OCR, accuracy is very dependent on source quality, but in all comparisons I have done we are basically on par with our competitors.

Price-wise... well, OCR with the (end user version) of our viewer is 100% free - even in the non licensed version.

And the PDF-XChange PRO SDK is very competitively priced. It includes our PDF creation and modification libraries, Print Drivers SDK, and Viewer SDKs (with limited included distribution, though for the Viewer SDK this initially includes 7500 end user desktop installs). Note also that we are essentially giving away the OCR functionality, since we only very recently added it to our already extensive SDK functionality, without changing the pricing in any way to recognize the significant value this adds.

-Walter

[aitchisj]

Thanks for the information. I'm still on your side... price-wise is the major seller for me... you blow the competition off the map in that department!

Now I just need to convince the bosses this is a good choice.
-John

[Paul - PDF-XChange]

Do let us know if you need anything from us here.

regards