How To Improve OCR Accuracy?

I spend the past few days on developing Android app to recognize English text on images to let the user copy/share text easily. I published a testing release of the app on Google Play.

The UI is awful but I want to improve accuracy of the app first. I am thinking of two methods:

1. improve tesseract traineddata file for better recognition of text.
2. use a false recognition dictionary which replace a word like boxmg to boxing as the in is false detected as m.

Let me know if you have any further improvements and suggestions for this Text OCR app.

THANKS FOR YOUR EFFORTS!

Comments

[Toma]

In my notes app play.google.com/store/apps/details... I also used tesseract API, I made it do the OCR in run-time and also added the turn the flash on/off feature. Haven't touched the app 3-4 years. Аlso I saw that Google also offers OCR: cloud.google.com/vision/docs/ocr but, of course, they try to lock us-devs into their cloud.