Skip to content

AI4Bharat/Shoonya

Repository files navigation


Shoonya

An open source platform to annotate and label data at scale

License: MIT


Shoonya is an open source platform to annotate and label data at scale, built with a vision to enhance digital presence of under-represented languages in India.

Shoonya offers support for multiple data types (Ex : parallel datasets, OCR, ASR, TTS etc) and labeling tasks (Ex : parallel datasets, OCR, ASR, TTS etc).

Shoonya, referring to zero, represents the start. It also represents universality in the sense that several cultures have a similar symbol for zero. We believe that the language resources that we collect in Shoonya will start a valuable open-source movement for Indian language technologies that will be available universally for all to adopt and improve upon.


Qualities of a good Data Collection Ecosystem

Challenges faced by Annotators

Why Shoonya?

The National Language Translation Mission (NLTM) has been announced in the budget by the Honorable Finance Minister in the backdrop of growing demand for accessing online services in local Indian languages. This will enable the wealth of governance-and-policy related knowledge on the Internet being made available in major Indian languages. The Ministry of Electronics and Information Technology (MeitY) has launched 'Bhashini' to help ensure that digital content is readily available to all citizens, in their preferred languages.

The goal of Bhashini is to develop an ecosystem of innovative practices for data collection, curation, develop technology for speech to speech translation and deliver solutions powered by open data, apps and services. Bhashini shall act as an orchestrator to bring contributions (like data, models etc.) received from government, industry, academia and society into an open “Hundi” or “Repository”. All contributions to Bhashini shall be validated and standardized using a Unified Language Contribution API (ULCA).

Reference : Bhashini Whitepaper

Also read Bhashini Data Report

Naturally, data collection/curation becomes the core of building state-of-the-art NLP ML models. This is where Shoonya comes into picture. Shoonya provides the platform for the Annotators/Translators to create such large datasets with highest quality.

Goals

  • Support all possible data types and labeling tasks
  • Build a reliable & scalable platform beneath Shoonya
  • Keep the UI simple and intuitive

Features of Shoonya

Overview and Demo Video

Shoonya Overview & Demo

Cloning this master repo

git clone --recurse-submodules https://github.com/AI4Bharat/Shoonya

Backend Setup

Clone the Shoonya-Backend repository from GitHub to your local machine.

git clone https://github.com/AI4Bharat/Shoonya-Backend.git

Create a virtual environment for the project. Replace with your preferred environment name.

python3 -m venv 

Activate the virtual environment. This ensures that the packages you install are isolated from the global Python environment.

source /bin/activate

Install all required Python packages listed in the requirements-dev.txt file.

pip install -r deploy/requirements-dev.txt

Set up the environment variables needed for the project by copying the example environment file.

cp .env.example ./backend/.env

Generate a new secret key for Django (within the virtual environment):

Open a Python shell.

python backend/manage.py shell

# Import the utility function to generate a secret key.
>> from django.core.management.utils import get_random_secret_key

# Generate and print a new secret key.
>> get_random_secret_key()

Copy the generated secret key and paste it into the .env file as the value for SECRET_KEY.

Docker Installation

Build the Docker containers as defined in the docker-compose-local.yml file.

docker-compose -f docker-compose-local.yml build

Run the containers in detached mode (-d flag). This will start up all the services defined in the Docker Compose file.

docker-compose -f docker-compose-local.yml up -d

Run Migrations

The following steps are required only when you run the project for the first time or after making changes to the models.

# Check if there are any pending migrations.
docker-compose exec web python backend/manage.py makemigrations 

# Apply all pending migrations to the database.
docker-compose exec web python backend/manage.py migrate

Create a superuser for accessing the Django admin interface (required only once).

docker-compose exec web python backend/manage.py createsuperuser

Run the Django development server within the Docker container.

docker-compose exec web python backend/manage.py runserver

Frontend Setup

Clone the Shoonya-Frontend repository from GitHub to your local machine.

git clone https://github.com/AI4Bharat/Shoonya-Frontend.git

Change directory to the newly cloned Shoonya-Frontend folder.

cd Shoonya-Frontend

Install the necessary dependencies for the project.

The --force flag is used to bypass conflicts with the existing dependencies.

npm i --force

Start the development server. This will run the frontend application on a local server.

npm start

Communication Forum

Any information/help/discussion required, can be taken up using the following link : https://github.com/AI4Bharat/Shoonya/discussions

Code of Conduct

This project adheres to the Contributor Covenant code of conduct. By participating, you are expected to uphold this code. Please report unacceptable behavior to [email protected].