DeepFake-Detection-Application Using Django

In recent years, deepfake technology has made huge leaps in terms of accuracy, quality, and most of all: believability. At one time or another, we've all been fooled by a computer-generated viral image or even someone's Snapchat filter. The ability to determine whether or not an image is reality is quickly becoming a necessity in a world that's becoming more and more susceptible to "questionable" information. Recently, doctored images have made headlines and divided people in their opinions of whether real-world issues were legitimate. In this project we have built a video classification system that can determine if it is real or deepfake. The propose system detects the DeepFake using Convolutional Neural Network (CNN) and Recurrent Neural Network (RNN).

In a narrow definition, deepfakes (stemming from “deep learning” and “fake”) are created by techniques that can superimpose face images of a target person onto a video of a source person to make a video of the target person doing or saying things the source person does. This constitutes a category of deepfakes, namely face swap. In a broader definition, deepfakes are artificial intelligence (AI)-synthesized content that can also fall into two other categories, i.e., lip-sync and puppet-master. Lip-sync deepfakes refer to videos that are modified to make the mouth movements consistent with an audio recording. Puppet-master deepfakes include videos of a target person (puppet) who is animated following the facial expressions, eye and head movements of another person (master) sitting in front of a camera

Requirements:

Note : Nvidia GPU is mandatory to run the application.

You can find the list of requirements in requirements.txt. Main requirements are listed below:

Python >= v3.6 Django >= v3.0

Directory Structure

• ml_app -> Directory containing code in views.py file
• project_settings -> Contains Django settings and files to run in production
• static -> Contains all css, js and json files (for face-api)
• templates -> Template files for HTML

Note: Before running the project make sure you have created directories namely models, uploaded_images, uploaded_videos in the project root and that you have proper permissions to access them.

Prerequisite

Copy your trained model to the models folder. You can download our trained models from the Google Drive or you can train your models using the steps mentioned in Model Creation directory.

Step 1 : Clone the repo and Navigate to Django Application

git clone https://github.com/Fatema110/DeepFake-Detection-Application.git

Step 2: Create virtualenv (optional)

python -m venv venv

Step 3: Activate virtualenv (optional)

venv\Scripts\activate

Step 4: Install requirements

pip install requirements.txt

Step 5: Copy Models

Copy your trained model to the models folder i.e Django Application/models/

You can download our trained models from Google Drive Note : The model name must be in specified format only i.e model_84_acc_10_frames_final_data.pt. Make sure that no of frames must be mentioned after certain 3 underscores _ , in the above example the model is for 10 frames.

Step 6: Run project

python manage.py runserver