0

I'm creating a python program that requires the use of an OCR since this program has to read and interpret Image text. I'm using EasyOCR for this project and it's been working quite well so far. The image gets processed before the OCR reads it, and it's the usual stuff like normalization, thresholding, text thinning, etc.

The text that's in these images are somewhat niche-ish? They include regular English words, but also codes that are used to identify them. Sadly these codes don't have a fixed format and so they're tricky to properly detect.

I was wondering what the benefits are of a custom OCR model? I'd like to know beforehand, just so I don't spend a while training my own model just for it to return the same results as the one I'm using currently.

In case it's helpful, here's a few examples of these codes I have to process:

V-40046-A

12411-A

W-90629

LC06853

6360 011

Thanks in advance!

Jack Coleman
  • 19
  • 2
  • The ostensible benefits would be the same as it’d be for “custom *anything*” else. You’d tailor the model to your specific use case for improved accuracy and performance – Paul H Jul 22 '23 at 14:06

0 Answers0