SignBridge 🤝

Real-Time Sign Language Recognition and Translation

📌 Overview

SignBridge is a real-time Sign Language Recognition and Translation system designed to bridge the communication gap between the Deaf/Hard-of-Hearing and the hearing community. It uses computer vision and deep learning to recognize ASL gestures and convert them into readable English text in real time.

🎯 Features

• 🔠 Real-time recognition of ASL alphabets (A–Z)
• 📷 Live webcam input and gesture detection
• 🧠 Lightweight MobileNetV2 model with high accuracy
• ✋ Hand landmark tracking using MediaPipe
• 💬 Dynamic translation display and reset functionality
• 🕒 Translation history tracking

🛠️ Tech Stack

• Frontend: HTML5, CSS3, JavaScript
• Backend: Python, Flask
• Machine Learning: TensorFlow/Keras (MobileNetV2)
• Computer Vision: OpenCV, MediaPipe

💻 Installation and Setup

1. Clone the repository

git clone https://github.com/kshitideshpande/SignBridge.git
cd SignBridge

2. Create virtual environment

python -m venv venv
source venv/bin/activate      # On Windows: venv\Scripts\activate

3. Install the dependencies

pip install -r requirements.txt

4. Unzip the trained model

• The trained model is included in the repository as a .7z file.
• Download and extract asl_mobilenetv2_mediapipe_resized.7z using 7-Zip or any compatible tool.
• Place the extracted asl_mobilenetv2_mediapipe_resized.h5 file in the project root directory.

5. Run the application

python app.py