🖐️ Sign Language Detection

A deep learning-based Sign Language Detection project that classifies American Sign Language (ASL) hand gestures.

🚀 Live Demo

🔗 Sign Language Detection App

---

📌 Features

• 🏗 CNN Model trained on ASL dataset
• 📷 Real-time gesture recognition
• 🖼 Image-based classification
• 📊 Data preprocessing and augmentation
• 🌍 Deployed using Streamlit

---

📥 How to Run Locally

1️⃣ Clone the Repository

git clone https://github.com/AryanDhanuka10/Sign_Language_Detection.git
cd Sign_Language_Detection

2️⃣ Setup Virtual Environment

pip install virtualenv
virtualenv venv
source venv/bin/activate   # On Windows, use: venv\Scripts\activate

3️⃣ Install Dependencies

pip install -r requirements.txt

4️⃣ Run the Application

streamlit run app.py

---

📊 Dataset

The dataset used is an American Sign Language (ASL) alphabet dataset (excluding the letters 'J' and 'Z' as they require motion).

📌 Dataset Source: Available on Kaggle

🖼 Sample Image:

---

🛠 Model Architecture

The project is built using Convolutional Neural Networks (CNNs) to classify ASL hand gestures.

🔹 Layers Used:

• Conv2D: Extracts spatial features from images.
• BatchNormalization: Normalizes activations to improve training.
• MaxPooling2D: Reduces feature dimensions while preserving important information.
• Dropout: Prevents overfitting by randomly deactivating neurons.
• Flatten: Converts multidimensional tensors into vectors.
• Dense: Fully connected layers for classification.

📌 Model Visualization:

---

🚀 Deployment

The project is deployed on Streamlit. You can access the live demo here:

🌐 Sign Language Detection App

---

🎯 Future Improvements

• 🎥 Add real-time video detection
• 🤖 Enhance model accuracy with more training data
• 📊 Implement Transfer Learning using pre-trained models
• 📱 Deploy as a mobile app

---

🤝 Contributing

Contributions are welcome! If you find a bug or want to improve the model, feel free to submit a Pull Request.

---

📜 License

This project is open-source and available under the MIT License.