Skip to content

This repository provides a deepfake detection toolkit leveraging CNNs and transfer learning. It guides you through building, training, and evaluating deepfake detection models. The code offers flexibility for customization and exploration, allowing you to experiment with various techniques.

License

Notifications You must be signed in to change notification settings

harsharajkumar/Deepfake-Detection

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

4 Commits
 
 
 
 
 
 
 
 

Repository files navigation

#Deepfake Detection using Convolutional Neural Networks (CNNs) and Transfer Learning

This repository implements deepfake detection models using Convolutional Neural Networks (CNNs) and transfer learning with the Xception architecture. It provides a framework for training and evaluating deepfake detection models on image and video datasets.

Key Features:

Custom CNN Architecture: Build and train a CNN model tailored for deepfake detection. Transfer Learning with Xception: Leverage pre-trained Xception weights for improved performance. Data Augmentation: Increase training data size and improve model generalization. Evaluation Metrics: Assess model performance using accuracy, F1-score, confusion matrix, ROC curve, and AUC. LIME Explanation (commented out): Gain insights into the model's decision-making process (requires additional setup). Video Processing (commented out): Extract frames, resize, and process video data for deepfake detection. Requirements:

TensorFlow Keras OpenCV Pandas NumPy (and other dependencies specified in requirements.txt) Installation:

Clone this repository: git clone https://github.com/your-username/deepfake-detection.git Navigate to the project directory: cd deepfake-detection Install required libraries: pip install -r requirements.txt Usage:

Prepare your deepfake dataset: Ensure your dataset is organized with video files and corresponding labels (real or fake) in a clear structure. Adjust data paths: Update the code (data_path variable) to point to your dataset location. Train the model: Run the script (python train.py). This script performs data loading, preprocessing, model training, evaluation, and saving. Evaluate the model: Analyze the generated metrics (accuracy, F1-score, confusion matrix, ROC curve, AUC) to understand the model's performance. Further Exploration:

Experiment with different hyperparameters (learning rate, number of epochs, etc.) to potentially improve performance. Explore other deep learning architectures suitable for deepfake detection. Implement video processing functionality for deepfake detection in videos.

Contributing:

We welcome contributions to this project! Please create a pull request to share your improvements.

About

This repository provides a deepfake detection toolkit leveraging CNNs and transfer learning. It guides you through building, training, and evaluating deepfake detection models. The code offers flexibility for customization and exploration, allowing you to experiment with various techniques.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published