Deepfake recognition

With the advent of deepfake and similar image forgery technology, it is becoming more and more difficult to differentiate between forgery and real media. Fortunately, just as neural networks can compose fake media, they can also detect it. In this recipe, we will utilize a deep neural network to detect fake images. The recipe utilizes the MesoNet architecture, found in the GitHub repository, DariusAf/MesoNet.

Getting ready

Preparation for this recipe consists of installing keras, tensorflow, and pillow in pip. The instructions are as follows:

pip install keras tensorflow pillow

In addition, a collection of fake and real images has been provided for you in the mesonet_test_images ...