OCR Text Scanner

Text OCR Scanner Software Tool. It is for extracting text from images and PDF files.

This software is UNDER HEAVY DEVELOPMENT. Use this OCR software for now.

commands

# get all dependencies of the project
go mod tidy

# build: produce an executable file/CLI app named "ocr"
go build -o ocr .

# run the app to identify a sample image
./ocr --lang=eng --img=img/default.png

# deps, build, run
go mod tidy && go build -o ocr . && ./ocr --lang=eng --img=img/default.png

# detect race conditions & memory leaks
go run -race .

Roadmap & Tasks

Done & TO-DO of features, goals and values. tiny steps towards the goal.

• v0.1.0
  • OCR library used is tesseract-OCR with the gosseract v2 wrapper.
  • image to text
  • support English OCR
  • support Arabic OCR
  • add tests
  • use some test images from renard314/textfairy
• Next
  • PDF to text
  • PDF to docx
  • PDF to selectable-text PDF
  • scalable : take advantage of all CPU cores to get the job done faster
  • bulk / patch-processing : coroutines and parallelism for tasks / jobs
  • composable CLI app for scripts and automation
  • UX / easy to use / user friendly
  • cut the image into pieces/segments and concurrently OCR them. (performance)
  • available on Debian & Debian-based distros
  • available as snap
  • available as flatpak
  • available on Elementary OS
  • available on Mac OS (likely via HomeBrew)
  • available on Windows
  • scan history : { original image, processed image, extracted text }
  • migrate from opencv into Go libs

References

• 3b1b : what is a convolution?
• train / refine tesseract OCR
• tesseractfonts : fine tune tesseract for new fonts
• Pre-Recognize Library - library with algorithms for improving OCR quality.

source code

The source code of OCR project can be found on:

• GitHub: https://github.com/abanoubha/ocr.git
• GitLab: https://gitlab.com/abanoubha/ocr.git
• CodeBerg: https://codeberg.org/abanoubha/ocr.git