Sign Language Recognition for Deaf and Dumb

• Our sign language recognition project involved creating a custom dataset, preprocessing images, training a model, integrating with React, and hosting with Firebase.

• The result is a real-time sign language recognition application that recognizes a variety of sign language gestures.

• Our Model is trained for 26 alphabets and 16 words of ASL and which are commonly used in general communication.

Features

• Real-Time Recognition

• Easy-to-Use Interface

• Adaptive Learning

• High Accuracy

• Real Time User Progress Data

Tech Stack

Front-end:

• React
• Redux

Back-end:

• Firebase (for hosting, authentication, and storage)

Machine Learning Framework:

• MediaPipe

NPM Packages:

• @mediapipe/drawing_utils
• @mediapipe/hands
• @mediapipe/tasks-vision
• @redux-devtools/extension
• chart.js
• firebase
• js-cookie
• react-chartjs-2
• react-icons
• react-redux
• react-router-dom
• react-toastify
• react-webcam
• redux
• redux-thunk
• uuid

Team Members

• So this project is a group project done in collaboration with the members mentioned below.

Name	Email-id
Shubham More	[email protected]
Sameer Metkar	[email protected]
Omkar Mandavkar	[email protected]
Durgesh Kolhe	[email protected]

Project Details

• Our Project Report: Report.pdf

• Published Paper: Paper.pdf

• Dataset Link: Sign_Dataset

• Gesture Recogition Documentation: Mediapipe

• The Model Training File is located in root folder

Steps to set up the project

• Check this document for project setup: SetUpDoc.docx

Authors

• @shubham-more
• @sameer-metkar
• @omkar-mandavkar
• @durgesh-kolhe

Acknowledgements

• React
• mediapipe
• Firebase
• NPM

Screenshots

Support

For support, contact

• email: [email protected]
• LinkedIn: Shubham More