Welcome to the Miraj deepfake detection repository! Our goal is to combat the proliferation of synthetic media by leveraging state-of-the-art deep learning techniques. DeepFake technology presents formidable challenges to media authenticity, and our system is designed to reliably detect manipulated content.
-
Deep Learning Models: We employ advanced deep learning models, including ResNext and LSTM, trained on extensive datasets to distinguish between genuine and manipulated media.
-
Multi-modal Analysis: Our system conducts comprehensive analysis across various modalities, including image, video, and audio, to ensure thorough assessment of media authenticity.
-
Real-time Detection: Capable of real-time detection, our system promptly identifies and flags potentially fraudulent content as it emerges.
-
Scalability: Designed with scalability in mind, our system can handle large volumes of media content across diverse platforms and applications.
To use our DeepFake Detection System, ensure you have the following dependencies installed:
- Python 3.x
- TensorFlow
- OpenCV
- NumPy
- SciPy
- Matplotlib
Check out the architecture diagram below to understand the components of our DeepFake Detection System:
We achieved impressive results with our model:
- Model Used: model_89_acc_40_frames_final_data.pt
- Number of Videos: 6000
- Number of Frames: 40
- Accuracy: 89.35%
Access the datasets used in our research:
Download our trained model from the following link:
We welcome contributions! Whether you have ideas for improvements, feature requests, or bug reports, feel free to submit an issue or create a pull request. For major changes, please open an issue first to discuss the proposed alterations.
Don't forget to ⭐ Star this repository to show your support!
Thank you for using Miraj deepfake detection!