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:
- improve tesseract traineddata file for better recognition of text.
- use a false recognition dictionary which replace a word like
boxmg
toboxing
as thein
is false detected asm
.
Let me know if you have any further improvements and suggestions for this Text OCR app.
THANKS FOR YOUR EFFORTS!
Top comments (1)
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.