Skip to content

color-theory/ocr-dataset-gen-node

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

16 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

simple-ocr-test

My first attempt at OCR.

It should take an image and parse out readable characters from top to bottom, left to right and return the value as a string.

Warning

This is a naive approach created for learning purposes. If you need a proper OCR library, it would be wise to look elsewhere.

Setup

You must first generate your reference vectors for each character. Add some ttf fonts to your fonts directory. Then

> npm run generate-vectors

You'll probably want to verify that they have been created and test properly against themselves. Expect one or two incorrect guesses.

> npm run test-vectors

Spell Checker

This project requires another project, the spell checker, to run by default. If you do not wish to use that project, there is an npm script that runs with the checker disabled. Otherwise, please download and run the server here before continuing.

https://github.com/color-theory/spell-check

Usage

> npm run ocr -- {Path to your image}

Or to run with spell checker disabled:

> npm run ocr -- --no-spell {Path to your image}

About

Using AI to recognize characters optically

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published