Real vs Fake Face Detector 🔍🤖

A machine learning-powered web app that detects whether a face image is Real or Fake (AI-generated). Built using a deep learning model served via a smooth Node.js + Express backend and styled with a simple frontend interface. This project is open to improvements and community contributions! 💫

🧠 How It Works

• User uploads a face image through the browser.
• The image is sent to the backend via AJAX (without page reload).
• Node.js server passes the image buffer to a Python script.
• A TensorFlow/Keras model processes the image and predicts Real or Fake.
• The prediction is returned as JSON and displayed live on the page.

🗃️ Dataset Used

This model was trained using the 140K Real and Fake Faces dataset from Kaggle.

📦 Dataset: https://www.kaggle.com/datasets/xhlulu/140k-real-and-fake-faces

⚠️ Note: This model performs best on images similar to the dataset (clean, centered, portrait-style images). Unseen fake styles or out-of-distribution faces may result in inaccurate predictions.

📦 Tech Stack

🚀 Getting Started

1. Clone the Repository

   git clone https://github.com/yourusername/RealVsFakeFaceDetector.git
   cd RealVsFakeFaceDetector

2. Install Node.js Dependencies

   npm install

3. Set Up Python Environment
   Make sure Python is installed (>= 3.10), and install required packages:

   pip install tensorflow numpy opencv-python

4. Start the Server

   node server.js

5. Visit the App in Your Browser

   http://localhost:4000

6. Repository link
   https://github.com/emaaanbutt/ReaVsFakeFaceDetector

🎯 Prediction UI

Users can upload an image and get results displayed live under the form — without any page reloads!

Example output:

Prediction: Real (99.34%)

Prediction: Fake (78.23%)

⚠️ Limitations

• 🚫 May misclassify GAN types not seen during training.
• 🧠 Trained on 140K images, mostly focused on a specific GAN type.
• 🖼️ Performance drops on heavily edited, noisy, or side-view face images.

✨ Future Improvements

• Add image preprocessing for better generalization.
• Enhance model with more GAN variety (StyleGAN2, etc.).
• Improve frontend with drag-and-drop uploads.

❤️ Acknowledgements

• Dataset: 140K Real and Fake Faces on Kaggle
• TensorFlow / Keras
• Node.js & Express
• jQuery for seamless AJAX form handling

📜 License

This project is licensed under the MIT License — free for personal or academic use. Please give credit if you use it or build upon it 💖

Preview