Data Augmentation for Convolutional Neural Network DeepFake Image Detection
Ameni Jellali, Ines Ben Fredj, Kaïs Ouni · 2023
We need to develop a technique for better identifying deepfakes because they can distort our perception of reality. This study offers a brand-new forensic technique for spotting falsified facial photos. We made advantage of the Kaggle- provided “real-and - fake- facial-detection” dataset. We are able to distinguish between probable facial alterations based on CNN's design. Thanks to data augmentation approaches, the results exhibit performances that are equivalent to those of previous works. The proposed approach fared better for this binary categorization into fake or real faces than the other cutting-edge studies. Our accuracy is close to 99 percent.