Image Authenticity Detector

Project Overview

The Image Authenticity Detector is a machine learning project designed to classify images as either real or AI-generated. This is done using a Generative Adversarial Network (GAN), which consists of two models: a generator and a discriminator. The discriminator is trained on a dataset containing both real and AI-generated images to enhance its ability to distinguish between the two.

The project uses Keras and TensorFlow to build the discriminator model, which classifies images as real (0) or fake (1).

Data Source

The dataset used for training the model is the CIFAKE dataset, sourced from Kaggle. It contains a collection of 60,000 synthetic images and 60,000 real images from CIFAR-10. The dataset is used to train the model on distinguishing between real and AI-generated images.

Dataset Link: CIFAKE: Real and AI-Generated Synthetic Images

Further details about the dataset can be found in the paper by Bird and Lotfi, 2024: CIFAKE: Image Classification and Explainable Identification of AI-Generated Synthetic Images.

Project Updates

This project uses the CCLE GAN model for the classification task, which is adapted into a binary classifier for detecting AI-generated images.

Installation Instructions

To set up the project, follow these steps:

1. Clone the repository:

   git clone https://github.com/Redgerd/Synthetic-Images-Detector
   cd Synthetic-Images-Detector

2. Install the required Python packages:

   pip install -r requirements.txt

3. Update the README to reflect any new changes or dependencies if necessary.

Usage

Once the dependencies are installed, you can start training the model or use the pre-trained model to classify images.

# Example: Load and use pre-trained model for classification
from keras.models import load_model
model = load_model('path_to_model')

# Assuming 'image' is a pre-processed image to classify
prediction = model.predict(image)
print("Real" if prediction == 0 else "AI-generated")

Links and References

• Papers with Code: CIFAKE on Papers with Code
• Original Paper: Bird, J.J. and Lotfi, A., 2024. CIFAKE: Image Classification and Explainable Identification of AI-Generated Synthetic Images. IEEE Access.
• Cycle GAN: Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks.