Detecção de deepfakes utilizando redes neurais convolucionais

Abstract

Exploring the use of convolutional neural networks (CNNs) to detect deepfake images and videos, media generated using machine learning techniques to manipulate or replace the original content. The study focuses on developing a CNN-based deepfake detection system that can accurately identify and distinguish between authentic and manipulated images and videos. The research involves training and testing different architectures using datasets of real and synthetic images and videos. The paper also discusses the impact of various factors on the performance of the CNN-based deepfake detection system, such as the type of deepfake generation method, the quality and resolution of the source media, and the size of the training dataset. The results demonstrate that CNN-based models can effectively identify deepfakes with high accuracy, and the proposed system can be applied in various real-world scenarios, including social media and journalism, to combat the spread of fake news and misinformation.