DEV Community

Abhishek Keshri
Abhishek Keshri

Posted on

Manyata - Recognize Hand Written Hindi Numerals


GitHub logo 2KAbhishek / Manyata

An OCR tool to recognize Hindi numerals. πŸ”ŸπŸ”Ž

An OCR tool to recognize Hindi numerals.



  • OS: Windows 7 or newer, Mac OS X 10.9 or newer, GNU/Linux, BSD and other Unix Systems.
  • Tools Used: Python, OpenCV, Tesseract


  • CPU: Intel Core 2 Duo or newer, AMD Phenom 2 X3 or newer
  • Memory: 4 GB or more
  • GPU: Any OpenGL 1.2 capable graphics card.
  • Storage: 4 GB or more


  • To identify hand-written Hindi numerals using neural networks.
  • To identify the numerals in an autonomous way.

Target Audience

Anyone trying to recognize hand-written Hindi numerals in an autonomous way.

Hello everyone, Abhishek here
Showing off my Project for #devgrad2020

How I built it

I have currently used Python, OpenCV, and Tesseract for building this project.
I got to learn a lot about OCR systems and also various things related to Computer Vision, like the LTSM models used in Tesseract 4.0 and above. Right now, I have added script which takes an example image as input and prints out the recognized characters on the terminal.


I'm also planning to work with tesseract.js and build a full featured web app in the future with what I have learned from this project.

Top comments (3)

mindset profile image
Jayesh Tembhekar ⚑

Hi, I like your project.
I also want to ask that, Can I apply for the Github Education Yearbook 2020?
I mean Is it necessary to be a final year student?
I am in 3rd year of engineering now πŸ€”

2kabhishek profile image
Abhishek Keshri • Edited

Thanks a lot Jayesh.
Yes, you have to be graduating this year in order to be part of the 2020 yearbook, you can check out the yearbook repo for more details.
Keep polishing your skills though, next year it'll be your turn.
Cheers πŸ˜„

mindset profile image
Jayesh Tembhekar ⚑

Thanks for reply πŸ™
I think I have to wait for next year πŸ˜…