🤟 SignScribe - Real-Time ASL Recognition

SignScribe is a machine learning-powered application that recognizes American Sign Language (ASL) signs in real-time using a webcam. It can detect the alphabet (A-Z) and special gestures like "Thank you" and "I Love You", providing a bridge between sign language users and technology.

---

🌟 Features

• ✅ Real-time ASL alphabet detection (A-Z)
• ✅ Special gesture detection: "Thank you" & "I Love You"
• ✅ Visual feedback with live webcam feed
• ✅ Gesture history display
• ✅ Stable predictions using motion tracking and history
• ✅ Built with Streamlit for a simple, interactive interface

---

🛠 How It Works

1. Data Collection

• train_model.py collects hand landmarks using MediaPipe.
• Users press keys (A-Z) to record gesture samples for the model.
• Data is saved in sign_data.csv for training.

2. Model Training

• Machine learning model reads CSV data and learns to predict signs.
• The trained model is saved as trained_asl_model.pkl for real-time detection.

3. Real-Time Detection

• detect_signs.py processes live webcam feed using MediaPipe Hands & FaceMesh.
• Hand landmarks are passed through the ML model to predict signs.
• Special motion logic detects:
  • "J": Pinky curves in motion
  • "Thank you": Hand touches chin and moves downward
  • "I Love You": Predefined static hand symbol
• Detected gestures are displayed on-screen along with recent history.

---

🎬 Installation

1. Clone the repository:

git clone https://github.com/Zaku-Seven/ASL_RECOGNITION.git
cd ASL_RECOGNITION

2. Create a virtual environment and activate it:

# Mac/Linux
python3 -m venv .venv
source .venv/bin/activate

# Windows
python3 -m venv .venv
.venv\Scripts\activate

3. Install dependencies:

pip install -r requirements.txt

4. Ensure trained_asl_model.pkl exists in the project directory, or generate it by running:

python train_model.py

---

🚀 Usage

1. Collect Training Data (Optional)

python train_model.py

• Position your hand inside the camera frame.
• Press A-Z to record gesture samples.
• Press Q to quit.

2. Run Real-Time Detection

python detect_signs.py

• Position your hand in the green detection box.
• Make clear ASL letters for recognition.

Special Gestures:

• J: Make "I" then curve your pinky down
• Thank You: Touch chin and move downward
• I Love You: Predefined static hand symbol

The camera feed will show detected signs along with a gesture history panel.

---

🗂 Project Structure

ASL_RECOGNITION/
│
├─ detect_signs.py        # Real-time detection & Streamlit interface
├─ train_model.py         # Train the ASL detection model
├─ sign_data.csv          # Collected hand landmark samples
├─ trained_asl_model.pkl  # Trained ML model
├─ requirements.txt       # Python dependencies
└─ README.md              # Project documentation

---

⚙️ Dependencies

• Python 3.11+
• OpenCV (cv2)
• MediaPipe
• NumPy
• Streamlit
• scikit-learn
• Pickle

ChatGPT and Claude.ai were briefly used to aid in machine learning