a-b-15/DCGAN

GAN Fake vs Real Classifier

This project is an end-to-end system to classify images as either "Real" or "Fake" (GAN-generated).

Project Structure

• train.py: Script to train the CNN model on the dataset.
• app.py: Flask web application for image classification.
• templates/: HTML templates for the web UI.
• Final Dataset/: Directory containing the training images.
• classifier_model.keras: The trained classification model (generated after training).

How to Run

1. Train the model:

   python train.py
   

2. Run the web app:

   python app.py
   

3. Open your browser and navigate to http://127.0.0.1:5000.

Architecture

The system uses a Convolutional Neural Network (CNN) built with TensorFlow/Keras. It includes:

• Rescaling layer (normalization)
• Multiple Conv2D and MaxPooling2D layers for feature extraction
• Dropout layer for regularization
• Dense layers for classification

UI Features

• Modern Glassmorphism design
• Drag and Drop image upload
• Instant classification results with confidence scores
• Responsive layout