Deep-Fake-Detection

#A Comparative Analysis of State-of-the-Art Algorithms for Robust Deep Fake Detection

You can access the relevant article from the repository: DeepFakeDetection.pdf

Project Overview:

In this Colab project, we conducted a comprehensive comparison of deepfake detection models, focusing on Convolutional Neural Networks (CNN), Gated Recurrent Units (GRU), and Long Short-Term Memory (LSTM) architectures. Our goal was to assess and contrast the performance of these models in identifying deepfake videos.

Experimental Setup:

• Platform: Google Colab
• GPU: NVIDIA T4
• Python Version: 3.10.12
• Libraries:
  • Torch (PyTorch) for deep learning
  • OpenCV (cv2) for video and image processing
  • Seaborn for visualization
  • Numpy for numerical operations

Results:

• CNN Model Accuracy: Approximately 60% on the DFDC dataset.
• LSTM Model Accuracy: Approximately 83% on the DFDC dataset.
• GRU Model Accuracy: Achieving an impressive accuracy of 85% on the DFDC dataset.