Sign Language Translator using MediaPipe, TensorFlow, and OpenCV

Overview Hey there! 👋 Welcome to the Sign Language Translator project. This application uses your webcam to recognize hand gestures in real-time and translates them into text. Whether it’s a static pose or a dynamic motion, this translator has got you covered!

It leverages:

• MediaPipe for precise hand landmark detection 🖐️
• TensorFlow for robust gesture classification 🧠
• OpenCV for seamless video processing 🎥
• CustomTkinter to make the interface smooth and simple

🚀 Features You’ll Love

• 👁️ Real-time gesture recognition using your webcam
• 🖖 Support for both static and dynamic gestures
• 💬 Translates gestures into text (words & sentences)
• 🎯 High accuracy with advanced normalization techniques
• 🖼️ User-friendly GUI built with CustomTkinter
• ⚡ Multiprocessing support for smooth performance

Tech Stack

• Python 🐍
• OpenCV 📸
• MediaPipe✋
• TensorFlow🧠
• NumPy 🔢
• CustomTkinter (CTk) 🖥️

🛠️ Get Started Ready to give it a spin? Follow these simple steps:

1. Clone the Repository

git clone https://github.com/your_username/sign-language-translator.git
cd sign-language-translator

2. Set Up Your Environment

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

3. Install Dependencies

pip install -r requirements.txt

4. Run the App Ensure your webcam is connected, then:

python main.py

5. Start Gesturing!

• Choose between live video or recorded video input.
• Perform sign language gestures in front of the camera.
• Watch your gestures transform into text on the screen!

---

##NOTE ALSO ENSURE THAT YOU HAVE UPDATE THE LOCATION OF IMAGE AND OTHER FILE LOCATION INORDER TO AVOID MISSING FILE ERROR 🧑‍🔬 How the Model Works The project uses a custom dataset of hand gestures, captured using MediaPipe's Hand Landmark Model. The dataset was labeled and used to train a neural network model (MLP or LSTM) in TensorFlow for accurate classification.

🔎 Normalization Techniques To improve accuracy, the model applies four key normalization techniques:

1. Wrist-Relative Normalization🏁

   • Aligns all landmarks relative to the wrist to remove positional bias.

2. Bone Length Normalization 🦴

   • Scales landmarks using the distance from the wrist to the middle finger MCP. This ensures hand size doesn’t affect recognition.

3. Z-Score Normalization📊

   • Standardizes landmark values to reduce noise and improve model performance.

4. Finger Spread Normalization ✋

   • Normalizes using fingertip distances, making gestures more comparable.

---

Contact Have questions or suggestions? Reach out via drop me an email at nithiishdanasekar@gmail.com

FINAL NOTE: THIIS PROJECT IS IN DEVELOPMENT BUT IT WORKS ONLY FOR STATIC GESTURES BUT NOT FOR DYNAMIC GESTURES please contact me for any QUERIES i AM ALWAYS AVAILABLE!!! Happy Coding! 💻✨