Final Year Project

MOCR (Multilingual Optical Character Recognition)

The project supports English and Devanagari characters.

View Demo · Documentation · Report Bug · Request Feature

📔 Table of Contents

• About the Project
  • [Datasets] (#datasets)
  • Screenshots
  • Tech Stack
  • Features
  • Color Reference
  • Environment Variables
• Getting Started
  • Prerequisites
  • Installation
  • Running Tests
  • Run Locally
  • Deployment
• Usage
• Roadmap
• Contributing
  • Code of Conduct
• FAQ
• License
• Contact
• Acknowledgements

🌟 About the Project

I have used this project using CRNN+Ctc loss. The OCR project has two languages: English and Devanagari. It uses Chain approximation and Otsu method for edge detection, and makes predictions based on the detected edges.

:datasets: About the datasets

For the English language model, I used the IAM English handwritten dataset. This is a widely recognized dataset for training and evaluating handwritten text recognition systems. To train the Devanagari model, I manually collected and labeled a dataset of Bhagwat Gita scriptures. Bhagwat Gita is a sacred Hindu text written in the Devanagari, Hindi with some sort of sanskrit script, providing a challenging and domain-specific dataset for the Devanagari OCR task.

📷 Screenshots

👾 Tech Stack

Modeling stacks

• MLTU
• Tensorflow
• Keras
• OpenCV
• Tensorboard

Server

• Python
• Django Rest Framework
• Cloudinery

Database

• SQLite

Client

• Vannila Javascript
• Html
• CSS
• Bootstrap

DevOps

• Docker

🎯 Features

• OCR model develop from ownselves instead of using Others
• Detect, Extract textual content from Image and export in desired format
• Supports Two language script, English and Devanagari

🎨 Color Reference

Color	Hex
Primary Color	#222831
Secondary Color	#393E46
Accent Color	#00ADB5
Text Color	#EEEEEE

🔑 Environment Variables

To run this project, you will need to add the following environment variables to your .env file

Cloudinery API_KEY for storing Uploaded image

Your email host password for recovering forgot password

🧰 Getting Started

‼️ Prerequisites

This project uses Docker for containerization. Make sure you have Docker installed on your machine. If not, you can download it from here.

🏃 Run Locally

Clone the project

  git clone https://github.com/AnisH1427/Multilingual-OCR-FYP.git

Go to the project directory

  cd Multilingual-OCR-FYP

Install pipenv if you haven't already

  pip install pipenv

Install dependencies

  pipenv install

Activate the pipenv shell

  pipenv shell

Start the server

  python manage.py runserver

🐳 Dockerization

Build the Docker image

docker build -t your-image-name .

Check the Docker images

docker images

Run the Docker container

docker run -d -p 8080:80 your-image-name

🚩 Deployment

To deploy this project, you can use the Docker container you built in the previous step.

🧪 Running Tests

To run tests, use the following command

  python manage.py test

🧭 Roadmap

• Data Acquisition
• Research
• Design Architecture
• Test with Different Hyperparameters
• Keep Training and Improving
• Design REST API
• Design User Interface
• Backend Setup
• Deploy Model
• Performance Monitoring Using Tensorboard

👋 Contributing

This project is Solely Contributed by MySelf as Final Year Project

📜 Code of Conduct

Please read the Code of Conduct

❔ Academic Questions and Answers

• What are the limitation in Current OCR so that Intelligent OCR is still the topic of research?

  • Answer 1

• Why OCR sysytem dont have different level of Performance in different languages?

  • Answer 2

• What can be done to improve the existing OCR systems?

  • Answer 3

⚠️ License

Distributed under MIT license

🤝 Contact

Your Name - @linkedin - anishkhatioda@outlook.com, anishkhatioda@gmail.com

Project Link: (https://github.com/AnisH1427/Multilingual-OCR-FYP)

💎 Acknowledgements

Use this section to mention useful resources and libraries that you have used in your projects.

• [Biru Shrestha] - Project Supervisor
• [Uttam Acharya] - Reader