Indian Sign Language (ISL) Detection System

Project Overview

The ISL Detection System is a web-based application that uses machine learning to recognize and classify Indian Sign Language (ISL) hand signs in real-time using webcam or image upload functionality.

Key Features

• Real-time webcam-based sign language detection
• Image upload for sign language recognition
• Machine learning model powered by Teachable Machine
• Responsive and user-friendly web interface
• Instant prediction with confidence percentage

Technologies Used

• HTML5
• CSS3
• JavaScript
• TensorFlow.js
• Teachable Machine
• MobileNet Architecture

Model Details

• Model Type: Image Classification
• Platform: Teachable Machine
• Base Architecture: MobileNet
• Dataset: Indian Sign Language Dataset

About MobileNet

MobileNets are lightweight, efficient neural network architectures designed for mobile and embedded vision applications. Key characteristics include:

• Small model size
• Low latency
• Low power consumption
• Balanced trade-off between accuracy and computational efficiency

Prerequisites

• Modern web browser
• Internet connection
• Webcam (for real-time detection)

How to Use

1. Open the web application
2. Wait for the model to load
3. Choose detection method:
   • Click "Start Webcam" for real-time detection
   • Upload an image for sign recognition
4. View instant prediction results with confidence percentage

Installation

1. Clone the repository

   git clone https://github.com/sumit-kumar-samal/ISL-Detection-System.git
   

2. Open index.html in a web browser

Technologies and Libraries

• TensorFlow.js
• Teachable Machine
• MobileNet

Limitations

• Requires modern browser support
• Performance depends on image quality and lighting
• Limited to trained sign language categories

Acknowledgments

• Teachable Machine
• TensorFlow.js
• Open-source community