💡 U.N.I.T.E. — Unified Node for Interpreting and Translating Expressions

Empowering the differently-abled through inclusive communication technologies

---

📖 Overview

U.N.I.T.E. is a multi-phase accessibility tool designed to bridge the communication gap between visually impaired and hearing impaired individuals and the rest of the world.

In a world where information access and communication are critical, UNITE addresses the pain points of:

• The slow, bureaucratic process of Braille document generation,
• The inaccessibility of sign language to non-signers,
• And the exclusion of visually impaired individuals from fast, digital communication.

This unified platform supports:

• 🔡 Text to Braille
• 🧏 Sign Language to Text
• 🧏‍♂️ Sign Language to Braille
• 🔊 Text to Speech
• 🧾 Audio to Braille
• ✍️ (Coming soon) Handwritten Text to All Phases

---

🚀 Key Features

• Instant Braille Conversion: Skip long wait times and bureaucracy. Input text, get tactile output.
• Sign Language Recognition: Convert hand gestures into real-time text using computer vision and ML.
• Speech Output: Converts detected text to speech for seamless audio interaction.
• Arduino-Powered Braille Printer: A compact, 6-servo module that embosses Braille in real-time.
• Custom Model Training: Train your own dataset with a few commands.

---

📦 Project Structure

UNITE/
├── Code/
│   ├── collect_imgs.py         # Captures 100 hand gesture images
│   ├── create_dataset.py       # Prepares dataset (coordinates to arrays)
│   ├── train_classifier.py     # Trains model using RandomForest
│   ├── signlang.py             # Sign recognition interface (main entry point)
├── Arduino/
│   └── UNITE_Braille_Bot.ino   # Arduino UNO control sketch for Braille bot
├── media/                      # Project images and demo references
├── README.md                   # This file
└── requirements.txt            # Python dependencies

---

🔧 Setup Instructions

🧠 Software Requirements

Install dependencies with:

pip install -r requirements.txt

Libraries used:

• opencv-python
• mediapipe
• pyfirmata
• gtts
• playsound
• scikit-learn
• numpy

🦾 Arduino Braille Bot

The UNITE Braille Bot runs on:

• Arduino UNO
• 6x SG90 Micro Servo motors
• Powered by a 9V battery

🔌 Circuit Details

• Servo VCC ➝ Breadboard +ve rail
• Servo GND ➝ Breadboard -ve rail
• Servo Signal ➝ Arduino digital pins: 8–13
• Battery GND ➝ Breadboard -ve rail
• Breadboard -ve ➝ Arduino GND (important!)

📌 Pins can be reconfigured in signlang.py (lines 12–17)

---

🧪 How to Run

🖐 From Scratch (Custom Dataset)

1. Capture Hand Gestures Run the following to capture 100 samples per gesture:

   python collect_imgs.py

2. Create Dataset

   python create_dataset.py

3. Train the Classifier

   python train_classifier.py

4. Start the Recognizer Interface

   python signlang.py

   This opens the webcam interface. Show hand gestures, and the system will:

   • Detect sign language
   • Convert to text
   • Speak the word (via TTS)
   • Send character commands to Braille bot via Arduino

---

🧠 Under the Hood

• MediaPipe: For real-time hand landmark detection.
• OpenCV: Camera handling and gesture capture.
• Random Forest: Fast, reliable classification for hand gestures.
• Arduino + PyFirmata: Communicates gestures to servo motors.
• gTTS + playsound: Converts recognized text into speech audio.

---

🧭 Roadmap / Future Phases

• ✍️ Handwritten Text Detection (In Progress)
• 📱 Android App Integration
• ☁️ Cloud Accessibility API
• 🧠 LLM Integration for Smart Text Conversion
• 📊 Dashboard for Visual and Accessibility Analytics

---

💡 Why It Matters

Accessibility should not be an afterthought — it's a right. UNITE provides:

• 🔓 Autonomy to differently-abled individuals
• 📚 Access to information at par with others
• 🔁 Bidirectional communication between the hearing, sighted, and differently abled
• ⚙️ Customizability for any language, region, or use-case

---

🤝 Contributions

Want to improve UNITE? Submit a pull request or open an issue.

All contributions that align with our mission of inclusion and accessibility are welcome.

🙏 Acknowledgements

Thanks to all open-source libraries and frameworks used in building this project. Special thanks to the accessibility community and mentors who helped shape the idea.

---