android-mrz-reader
android-mrz-reader copied to clipboard
mercuriete
mrz.traineddata and How to implement OCR-B font?
Can't find in the code how does it implement mrz.traineddata and what is it main role? Also Is there a way to implement OCR-B font that is used for passports, because I cant find anything about that.
Best Regards
Hi @nikolacrnicrni I had a lot of problems about traineddata.
You can check the documentation in this url:
https://github.com/tesseract-ocr/tesseract/wiki/Data-Files#format-of-traineddata-files
This project is a spare time project so I will update it on "best effort".
Anyway, in my investigations... the best performance is using generic trained data from english.
PS: for training tesseract i used this app -> https://www.youtube.com/watch?v=HgeksbqH0ng
No, at the end I used .so file made in C++ and implemented in android project to make scaning.
@nikolacrnicrni sorry if I couldn't help you.
I didn't have any time to check the code. If you want to share your findings about trained data here you are welcome.
Thanks for checking this repository :)
@nikolacrnicrni , can you please share details which .so file you used and was it working perfect?
@faiqmustaqeem I'm sorry but no. I am now selling that .so file because I lost months of developing. It is created with OpenCV.
@nikolacrnicrni Remember the license of the .so you are selling. Remember if you sell a opensource project you have to comply with the license and provide de source code.
@mercuriete Yeah I know. But I started the project from begining with my code with OpenCV library.
OpenCV was built to provide a common infrastructure for computer vision applications and to accelerate the use of machine perception in the commercial products. Being a BSD-licensed product, OpenCV makes it easy for businesses to utilize and modify the code.
https://github.com/opencv/opencv/blob/master/LICENSE
@nikolacrnicrni Nice! This is the awesome of good licenses! I hope you are doing well in your bussiness 👍
Do you had problems with zero "0" and uppercase "o"?
I remembered some time ago trying to read European Visas, there was a field outside the checksum that was difficult to read with any computer vision framework. So in the end we fixed by hand because we knew it should be a number.