DeepFakes: Detecting Forged and Synthetic Media Content Using Machine Learning

View on arXiv ← Back to list

Authors: Sm Zobaed, Md Fazle Rabby, Md Istiaq Hossain, Ekram Hossain, Sazib Hasan, Asif Karim, Khan Md. Hasib

Published: 2021-09-07 05:19:36+00:00

AI Summary

This research paper reviews existing challenges, research trends, and future directions in DeepFake creation and detection techniques. It analyzes various DeepFake generation approaches (complete face synthesis, identity swap, attribute manipulation, face reenactment) and detection methods (forensics-based, deep neural network-based, GAN redesign-based, visual and audio inconsistency-based).

Abstract

The rapid advancement in deep learning makes the differentiation of authentic and manipulated facial images and video clips unprecedentedly harder. The underlying technology of manipulating facial appearances through deep generative approaches, enunciated as DeepFake that have emerged recently by promoting a vast number of malicious face manipulation applications. Subsequently, the need of other sort of techniques that can assess the integrity of digital visual content is indisputable to reduce the impact of the creations of DeepFake. A large body of research that are performed on DeepFake creation and detection create a scope of pushing each other beyond the current status. This study presents challenges, research trends, and directions related to DeepFake creation and detection techniques by reviewing the notable research in the DeepFake domain to facilitate the development of more robust approaches that could deal with the more advance DeepFake in the future.

Key findings

The review reveals limitations in current DeepFake generation (low resolution, limited attribute manipulation, lack of video continuity) and detection (simple baselines, evaluation on limited datasets, lack of robustness to attacks). The authors emphasize the need for high-resolution image generation, more generalizable attribute manipulation, and robust, deployable DeepFake detectors.

Approach

The paper conducts a comprehensive review of existing literature on DeepFake generation and detection methods. It classifies approaches based on their methodologies and analyzes their strengths and weaknesses, highlighting challenges and opportunities for future research.

Datasets

CASIA-WebFace, CelebA, VGGFace, MS-Celeb-1M, VGGFace2, Flickr-Faces-HQ, DeepFake Detection Challenge (DFDC) Preview Dataset, Celeb-DF Dataset, FaceForensics++, UADFV dataset, TIMIT

Model(s)

Various deep learning models are mentioned in relation to existing work, including GANs, VAEs, autoencoders, CNNs, RNNs, LSTMs, and XceptionNet. Specific model architectures within these categories are also detailed, though not a single model is used as the main contribution of the paper.

Author countries

USA, USA, USA, Australia, USA, Australia, Bangladesh

← Previous