Training OCR

Report · Dec 12, 2017

How do I train the OCR in Adobe Acrobat Pro DC to stop making the same mistake over and over? About half the time, when a document has "1." (one period) the OCR sees "1 ." (one space period). It's critical that it get the numbers right. This is about the only mistake it makes. The number is crystal clear on the original. Adobe doesn't think it's made a mistake and doesn't ask for my interpretation of the number in OCR. Adobe is worthless to me in this case if I can't train it to correctly interpret the number. I'm scanning a high resolution and have used the enhancement feature. Thanks in advance for your help.

Report · Dec 12, 2017

Acrobat’s OCR engine cannot be trained. It is what it is. OCR in Acrobat is good and convenient, but a dedicated OCR solution will always be better. I use Abbyy’s FineReader for more demanding OCR jobs, and the built in Acrobat OCR for anything more run of the mill.

Report · Dec 13, 2017

Thanks, I'll try Abby's. It's odd that if Acrobat doesn't know if it got it right, it will allow you to train it. For example, The OCR had trouble with some handwritten notes at the top of the document. It asked me to verify or correct it's OCR through training. From then on, it got them right. So there must be some training in the program. On characters it thinks it got right, but didn't, you can't train. It will never get them right. It keeps making the same mistake from now on. I have to say that the it's pretty arrogant for anyone to think they can write a program that never needs correcting. Again, the document is clean and I've done all the enhancement I can. Adobe is only half a step away from allowing the users to tell it that it made a mistake and train it. Again, I'll look at Abby. Thanks very much for the good advice.