Knowledge digitization plays a key role in the adaptive learning system that I am building. Digitization is a time consuming process, where the majority of time is bled in transferring knowledge from physical to digital medium - not just the as-is content (OCR), but transformation of the text into intelligent questions like question/answer, fill in the blanks, true/false, definitions, word meaning etc.
I have been working on a process optimization strategy to accelerate the digitization process. A key ask of the optimized workflow was that of a super fast scanner - how fast? well it should be able to scan a 300 page book in say 10 minutes - with OCR quality resolution :)
An obvious pit stop in the research for do-it-yourself book scanning is http://www.diybookscanner.org/. Unfortunately, I found the designs too exorbitant and over-engineered for my purpose. Well, I have to mention that two of my severe constraints were cost (of course), but more importantly the limitations of my modest workshop and the availability of raw materials.
And so, I had to create my own design for the scanning hardware (blueprint below). Proud to say that it is working perfectly. The best part is, it works perfectly with my cell phone camera (Samsung Galaxy S4).
The gist for the image processing script is here:
https://gist.github.com/deb-sandeep/e2c30129fbb346e8a809#file-process-scan-sh
After image processing ...
No comments:
Post a Comment