VideoDeepFakeDetection

Application that detects the originality of video files with artificial intelligence.

Setup Environment

# Make sure your PIP is up to date
pip install -U pip wheel setuptools

# Install required dependencies
pip install -r requirements.txt

Application

• You can run the file named main.py.
• Running on http://127.0.0.1:5000

• Load your video(.mp4) file and test whether the file is real or not.

Overview

1- The video file is opened, and various video properties such as fps, width, and height are obtained.

2- Face detection is performed using MTCNN (Multi-Task Cascaded Convolutional Networks).

3- The detected face is transformed into a feature vector using a pre-trained Inception Resnet V1 model (InceptionResnetV1).

4- A comparison is made with the face in the previous frame, and a similarity score is calculated.

5- Similarity scores below a certain threshold are considered as indicative of a deepfake.

6- If deepfakes are detected in a consecutive number of frames, it is marked as a deepfake, and a frame is added to the video.

7- Processed frames are written to an output video file.

Contributing

If you want to contribute to this project, please follow these steps:

• Fork: Fork this repository to your GitHub account.
• Create a Branch: Create a new branch to add a new feature or fix a bug.
• Commit: Add clear commit messages explaining your changes.
• Push: Push your changes to the repository you forked.
• Pull Request: Create a pull request on GitHub.

License

Our project is licensed under the MIT License.