Home > Software design >  Tesseract OCR or OpenCV for non-existent language
Tesseract OCR or OpenCV for non-existent language

Time:03-17

There is an alphabet (https://i.stack.imgur.com/txrYB.jpg) that matches each character with an English letter, I also have SVG images for each character. I want to write a program that, based on a screenshot (like this: https://i.stack.imgur.com/8sY2v.jpg), will "translate" the text from this screenshot to English for me. Can you tell me what I need to use to write such a program?

CodePudding user response:

To do that, you need to "train" Tesseract to understand your alphabet. It's not trivial.

https://tesseract-ocr.github.io/tessdoc/tess4/TrainingTesseract-4.00.html

CodePudding user response:

This means you need to get an OCR engine that supports this language. Even if you train Tesseract, you train it to support a new font for an existing language, not a new one.

  • Related