Deep Learning for Deepfakes Creation and Detection: A Survey

View on arXiv ← Back to list

Authors: Thanh Thi Nguyen, Quoc Viet Hung Nguyen, Dung Tien Nguyen, Duc Thanh Nguyen, Thien Huynh-The, Saeid Nahavandi, Thanh Tam Nguyen, Quoc-Viet Pham, Cuong M. Nguyen

Published: 2019-09-25 16:03:45+00:00

AI Summary

This research paper surveys deep learning algorithms used for creating and detecting deepfakes. It provides a comprehensive overview of existing methods and discusses challenges, research trends, and future directions in this field.

Abstract

Deep learning has been successfully applied to solve various complex problems ranging from big data analytics to computer vision and human-level control. Deep learning advances however have also been employed to create software that can cause threats to privacy, democracy and national security. One of those deep learning-powered applications recently emerged is deepfake. Deepfake algorithms can create fake images and videos that humans cannot distinguish them from authentic ones. The proposal of technologies that can automatically detect and assess the integrity of digital visual media is therefore indispensable. This paper presents a survey of algorithms used to create deepfakes and, more importantly, methods proposed to detect deepfakes in the literature to date. We present extensive discussions on challenges, research trends and directions related to deepfake technologies. By reviewing the background of deepfakes and state-of-the-art deepfake detection methods, this study provides a comprehensive overview of deepfake techniques and facilitates the development of new and more robust methods to deal with the increasingly challenging deepfakes.

Key findings

The number of research papers on deepfakes has increased significantly in recent years. Current deepfake detection methods face challenges regarding generalization to unseen forgeries and cross-dataset scenarios. Future research should focus on creating larger benchmark datasets and developing more robust, explainable, and generalizable detection methods.

Approach

The paper conducts a survey of deepfake creation and detection methods, categorizing them based on data type (images or videos) and features used (handcrafted or deep). It analyzes existing techniques and highlights challenges and future research directions.

Datasets

The survey mentions numerous datasets used in various studies, including CelebA, FaceForensics++, DeepfakeDetection, Celeb-DF, Deepfake Detection Challenge (DFDC), DeeperForensics-1.0, VidTIMIT, Hollywood human actions dataset, UADFV, DeepfakeTIMIT, Idiap Research Institute replay-attack dataset, and ILSVRC12.

Model(s)

The survey covers various models, including autoencoders, Generative Adversarial Networks (GANs), specifically StyleGAN and PGGAN, CNNs (e.g., VGG, ResNet, MesoNet, XceptionNet), Recurrent Neural Networks (RNNs), LSTMs, capsule networks, and Siamese networks.

Author countries

Australia, Australia, Vietnam, Vietnam, Republic of Korea, Australia, Vietnam, Republic of Korea, France

← Previous