Blogs

The “Cloud” is not new, nor is it news.  However, I fear the terrible things that might come to me if I don’t sacrifice an article or two to the “Cloud” gods.  No the “Cloud” is not new; it’s been around since before I was born when mainframe computers were the norm.  The “Cloud” also really tells you nothing about a solution; it only tells you about its architecture, because “SaaS”, “Cloud”, “ASP” are all technology architecture terms not solution terms. However, what this architecture brings is a grab bag of new solutions.  Maybe the marketers hit the lotto with a ridiculous renaming of a term, but we are all benefiting from the cool new solutions being developed because of it.  One of those solutions is web-based OCR or dare I say “OCR in the Cloud”.

The first time I used OCR in the “Cloud” was six years ago when I was playing with an FTP based OCR service, that quickly transitioned to an email based OCR service, which quickly transitioned into a browser based OCR service.  I could even upload photos from my phone and view results, before the arrival of the true smart phones.  Server side OCR had already been very common for service bureaus and large volume processing environments, but distributed amongst the consumers it was not.  New web-based OCR solutions are now clearly targeted at the consumer by collecting and processing images from the cell phone via an App or in the browser via file upload.

What is great about these new applications is the user does not have to understand anything about OCR only their goals.  Part of the challenge in OCR adoption has been getting users to understand first what it was, and second how to use it.  Maybe our AIIM community is familiar with the OCR acronym, but the general populous, though beneficial to them, could care less.  Those users who did own and operate a desktop OCR clients had the problem of picking the right settings, and eventual gained accuracy rage.  Now with “Cloud” based services the technology can determine the proper switches and levers to push and pull in order to receive optimum results. Provided the user did not upload junk, the results will be good. Web based OCR in most cases is using the latest and greatest stuff with the proper settings.  And now users are leveraging OCR in many applications and not even knowing it, getting the benefit without the fuss.  Normally, I would advocate greater market understanding of technology, but in the case of OCR the consumer market is just unwilling. Improved user adoption is relying on ease-of-use, which the “Cloud” architecture offers.

That is the good of web based OCR, now the bad.  First, the privacy of your images. All OCR solution providers Love, Love, Love sample images.  Samples allow them to create industry specific demos, increase use case understanding, and even improve OCR engine results.  So when you upload your image to the web do you know where it is going?  The regulations behind this, with the exception of healthcare, have yet to be defined.  There is a chance your images are being kept for a period of time or indefinitely and re-used for other purposes.   Second bad, is integration.  From a full-page OCR perspective, the “Cloud” is the right architecture, and the approach is mature. From an advanced capture perspective, a lot has to be figured out.  One key aspect of advanced capture is the feedback loop from designers, verifiers, and processing stations.  This over the web will have to be an integration from multiple “Cloud” based services, or even desktop clients.  The risk is, if users are expecting data capture in the “Cloud” because they can so easily do full-page OCR, this might cause push back. However, this is also a very exciting opportunity for web based character level verification systems, web based template setup, crowd sourcing, etc. etc. 

In my particular environment, I use OCR as a service in several ways, on my phone via three separate applications, and on my own network as (are you ready for this cutting edge term) a private “Cloud”.  Though the concept is not really new, it’s very exciting to see its developments.  The quality of OCR from your mobile device is going to increase as more use cases and samples are gathered. And the prospect of full-page web-based document conversion branching into advanced capture, natural language processing NLP, and being a part of larger “Cloud” based systems is inevitable and very cool to think about.

So there you have it, my sacrifice.  Although I’m not impressed with “The Cloud” the hype has forced a lot of cool solutions out of the woodwork, and many more to come.