Samvaad Saathi: Your Sign Language Translator

This is a machine learning app for real-time Sign Language Detection and Translation, built using the YOLOv5 model and implemented using PyTorch and OpenCV libraries. It can translate common phrases in sign language such as 'hello', 'sorry', 'thank you', 'I love you', etc.

The static web application can be accessed at Streamlit App since Streamlit Cloud doesn't support webcam. For dynamic access, visit the Installation, Local-Usage and App-Usage sections.

Table of Contents

• Features
• Tech-Stack
• Screenshots
• Installation
• Local-Usage
• App-Usage
• Tip

Features

1. Real-time sign language detection and translation
2. Buttons for starting and stopping video capture
3. Common Sign Language phrases dictionary in the side bar

Tech-Stack

1. Streamlit Framework for Frontend
2. YOLOv5 as the base model for sign language detection
3. Roboflow for data annotation
4. Python modules:
   1. Pytorch for implementation of YOLOv5 model
   2. OpenCV for video capturing and data collection
   3. Weights and Biases (WandB) for model training
   4. OS for executing commands on the local machine
   5. And many more packages like sys, uuid, time, pathlib, argparse, yaml, IPython, etc.
5. VSCode and Google Colab for development requirements
6. Webcam and Windows OS for hardware requirements

Screenshots

Installation

Follow these steps to clone and set up the repository locally:

1. Clone the repository:

   git clone https://github.com/SemalJohari/Samvaad-Saathi.git
   

2. Navigate to the project directory:

   cd Samvaad-Saathi
   

3. Create and activate a virtual environment:

   python -m venv venv
   venv\Scripts\activate
   

4. Install the required dependencies:

   pip install -r requirements.txt
   

Local-Usage

To run the project locally, start the development server or application using:

python run.py

App-Usage

To run the project using a web application with a local host, start the development server or application using:

streamlit run app.py

Tip

For running the project on Windows, add the following lines of code immediately after importing libraries in app.py and detect.py to switch from the Posix Path (for MacOS) to Windows Path (for Windows OS) using the pathlib library:

temp = pathlib.PosixPath
pathlib.PosixPath = pathlib.WindowsPath