Quote:
Using OCR, I'm getting 100% recognition, even with different backgrounds. There are good approaches for scraping tables, not only "plain" OCR.
Hey that is pretty impressive! I could do better with my OCR error rate but it has not been enough of an issue for me to fix it. I am using the Tessaract engine which I hear is not that great. For numbers it works pretty well. I do have to blow up the image and convert it to black and white before running it thru however. I did have a problem in which the number 5 would come thru as the letter S but it was consistent and easy to fix.Statistics: Posted by shalako — Wed Jan 07, 2015 11:33 pm
]]>