DeepFakes Evolution: Analysis of Facial Regions and Fake Detection Performance
Media forensics has attracted a lot of attention in the last years in part due to the increasing concerns around DeepFakes. Since the initial DeepFake databases from the 1st $$\text {1}^{\text {st}}$$ generation such as UADFV and FaceForensics++ up to the latest databases of the 2nd $$\text {2}^{\te...
Saved in:
| Published in | Pattern Recognition. ICPR International Workshops and Challenges Vol. 12665; pp. 442 - 456 |
|---|---|
| Main Authors | , , , |
| Format | Book Chapter |
| Language | English |
| Published |
Switzerland
Springer International Publishing AG
2021
Springer International Publishing |
| Series | Lecture Notes in Computer Science |
| Subjects | |
| Online Access | Get full text |
| ISBN | 3030688208 9783030688202 |
| ISSN | 0302-9743 1611-3349 |
| DOI | 10.1007/978-3-030-68821-9_38 |
Cover
| Summary: | Media forensics has attracted a lot of attention in the last years in part due to the increasing concerns around DeepFakes. Since the initial DeepFake databases from the 1st $$\text {1}^{\text {st}}$$ generation such as UADFV and FaceForensics++ up to the latest databases of the 2nd $$\text {2}^{\text {nd}}$$ generation such as Celeb-DF and DFDC, many visual improvements have been carried out, making fake videos almost indistinguishable to the human eye. This study provides an exhaustive analysis of both 1st $$\text {1}^{\text {st}}$$ and 2nd $$\text {2}^{\text {nd}}$$ DeepFake generations in terms of facial regions and fake detection performance. Two different methods are considered in our experimental framework: i) the traditional one followed in the literature and based on selecting the entire face as input to the fake detection system, and ii) a novel approach based on the selection of specific facial regions as input to the fake detection system.
Among all the findings resulting from our experiments, we highlight the poor fake detection results achieved even by the strongest state-of-the-art fake detectors in the latest DeepFake databases of the 2nd $$\text {2}^{\text {nd}}$$ generation, with Equal Error Rate results ranging from 15% to 30%. These results remark the necessity of further research to develop more sophisticated fake detectors. |
|---|---|
| Bibliography: | Original Abstract: Media forensics has attracted a lot of attention in the last years in part due to the increasing concerns around DeepFakes. Since the initial DeepFake databases from the 1st\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\text {1}^{\text {st}}$$\end{document} generation such as UADFV and FaceForensics++ up to the latest databases of the 2nd\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\text {2}^{\text {nd}}$$\end{document} generation such as Celeb-DF and DFDC, many visual improvements have been carried out, making fake videos almost indistinguishable to the human eye. This study provides an exhaustive analysis of both 1st\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\text {1}^{\text {st}}$$\end{document} and 2nd\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\text {2}^{\text {nd}}$$\end{document} DeepFake generations in terms of facial regions and fake detection performance. Two different methods are considered in our experimental framework: i) the traditional one followed in the literature and based on selecting the entire face as input to the fake detection system, and ii) a novel approach based on the selection of specific facial regions as input to the fake detection system. Among all the findings resulting from our experiments, we highlight the poor fake detection results achieved even by the strongest state-of-the-art fake detectors in the latest DeepFake databases of the 2nd\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\text {2}^{\text {nd}}$$\end{document} generation, with Equal Error Rate results ranging from 15% to 30%. These results remark the necessity of further research to develop more sophisticated fake detectors. |
| ISBN: | 3030688208 9783030688202 |
| ISSN: | 0302-9743 1611-3349 |
| DOI: | 10.1007/978-3-030-68821-9_38 |