Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

A Review of Deep Learning-based Approaches for Deepfake Content Detection

Leandro A. Passos, Danilo Jodas, Kelton A. P. da Costa, Luis A. Souza Júnior, Douglas Rodrigues, Javier Del Ser, David Camacho, João Paulo Papa

TL;DR

This survey addresses the growing problem of deepfake content by systematically reviewing deep learning-based detection methods across CNNs, generative-models, RNNs, and Transformers. It highlights a taxonomy of spatial versus temporal detection, catalogs major datasets (e.g., FaceForensics++, DFDC, Celeb-DF, WildDeepfake), and summarizes representative methods and their performance. The paper also discusses open issues such as cross-dataset generalization, the need for multimodal and explainable approaches, and future directions including unsupervised learning and diffusion-model-generated content. Its findings underscore the practical importance of robust detection systems for safeguarding digital media integrity amid rapidly evolving generative technologies.

Abstract

Recent advancements in deep learning generative models have raised concerns as they can create highly convincing counterfeit images and videos. This poses a threat to people's integrity and can lead to social instability. To address this issue, there is a pressing need to develop new computational models that can efficiently detect forged content and alert users to potential image and video manipulations. This paper presents a comprehensive review of recent studies for deepfake content detection using deep learning-based approaches. We aim to broaden the state-of-the-art research by systematically reviewing the different categories of fake content detection. Furthermore, we report the advantages and drawbacks of the examined works, and prescribe several future directions towards the issues and shortcomings still unsolved on deepfake detection.

A Review of Deep Learning-based Approaches for Deepfake Content Detection

TL;DR

This survey addresses the growing problem of deepfake content by systematically reviewing deep learning-based detection methods across CNNs, generative-models, RNNs, and Transformers. It highlights a taxonomy of spatial versus temporal detection, catalogs major datasets (e.g., FaceForensics++, DFDC, Celeb-DF, WildDeepfake), and summarizes representative methods and their performance. The paper also discusses open issues such as cross-dataset generalization, the need for multimodal and explainable approaches, and future directions including unsupervised learning and diffusion-model-generated content. Its findings underscore the practical importance of robust detection systems for safeguarding digital media integrity amid rapidly evolving generative technologies.

Abstract

Recent advancements in deep learning generative models have raised concerns as they can create highly convincing counterfeit images and videos. This poses a threat to people's integrity and can lead to social instability. To address this issue, there is a pressing need to develop new computational models that can efficiently detect forged content and alert users to potential image and video manipulations. This paper presents a comprehensive review of recent studies for deepfake content detection using deep learning-based approaches. We aim to broaden the state-of-the-art research by systematically reviewing the different categories of fake content detection. Furthermore, we report the advantages and drawbacks of the examined works, and prescribe several future directions towards the issues and shortcomings still unsolved on deepfake detection.
Paper Structure (31 sections, 15 figures, 2 tables)

This paper contains 31 sections, 15 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Review methodology
  3. Search Keywords
  4. Research Databases
  5. Selection Criteria
  6. Selected studies
  7. Deepfake Detection Methods Review
  8. Convolutional Neural Networks
  9. Generative Models
  10. Recurrent Neural Networks
  11. LSTM
  12. GRU
  13. Transformers
  14. Datasets
  15. HOHA-based dataset
  16. ...and 16 more sections

Figures (15)

  • Figure 1: Proposed methodology for the literature review.
  • Figure 2: Taxonomy of the deepfake detection methods.
  • Figure 3: General strategy for the learning of temporal sequences of $t$ frames in videos.
  • Figure 4: Samples from Faceswap-GAN dataset. For each block, the left column denotes original images and the right column stands for synthetic instances. Adapted from faceswapGAN.
  • Figure 5: Sample frames from the UADFV dataset. The top row depicts original faces, while the bottom row stands for synthetic images. Adapted from li2018ictu.
  • ...and 10 more figures