Deepfake Detection Model

This project implements a Cross-Attention Fusion Network for deepfake detection, combining EfficientNet-B2 (Spatial) and ResNet18 (Frequency) features.

📋 Prerequisites

• Python: 3.10 or higher recommended.
• CUDA-capable GPU: Highly recommended for training (Model is configured for cuda).
• RAM: At least 16GB recommended.

🛠️ Environment Setup

Follow these steps to set up the development environment using either venv or Conda.

1. Create Environment

Option A: Using venv (Standard Python)

# Windows
python -m venv venv
.\venv\Scripts\activate

# Linux/Mac
python3 -m venv venv
source venv/bin/activate

Option B: Using Conda

conda create -n deepfake-detection python=3.10
conda activate deepfake-detection

2. Install Dependencies

Install all required packages from the requirements.txt file.

pip install -r requirements.txt

3. Verify Installation

Ensure that PyTorch can see your GPU:

import torch
print(torch.cuda.is_available()) 

📂 Project Structure

• notebooks/: Contains the training notebook (model.ipynb) and experiments.
• data/: Dataset directory (expected structure below).

📊 Dataset Setup

The training notebook expects the data to be organized as follows:

data/
└── faces/
    ├── real/           # Real video files
    └── fake/
        └── Deepfakes/  # Deepfake video files

The script will automatically handle frame extraction to data/faces/extracted_frames/.

🚀 Running the Training

1. Open the notebook:

   jupyter notebook notebooks/model.ipynb

2. Run all cells to process data, extract frames, and train the model.