A Novel Approach for Detecting Deepfake Face Using Machine Learning Algorithms

In today's digital age, the ability to identify, differentiate, and authenticate manipulated online content is essential. Being ability to discriminate between the real and the fake is crucial. Recent advances in technologies such as artificial intelligence, machine learning, and deep learning are playing a major role in the generation of deepfake media (images and videos). Very realistic deep fake images and videos can be produced by utilizing sophisticated deep learning models such as generative adversarial neural networks (GAN s) and autoencoders, in conjunction with a sizable image collection pertaining to the subject matter. Deepfakes (DF) refer to artificially synthesized images or videos created using features such as face swapping and facial expression recombination. These face manipulation techniques have become extremely sophisticated. Deepfakes can be used to create child pornography, pornographic images of celebrities, revenge porn, fake news and harassment, spreading disinformation on social media platforms, financial fraud, election manipulation, and more. Therefore, there is a need to design and develop a robust framework to identify these deepfake images and videos. The purpose of this paper is to identify deepfakes from visual deepfake datasets and perform a comparative analysis of deep fake detection through machine learning algorithms.