• 文献检索
  • 文档翻译
  • 深度研究
  • 学术资讯
  • Suppr Zotero 插件Zotero 插件
  • 邀请有礼
  • 套餐&价格
  • 历史记录
应用&插件
Suppr Zotero 插件Zotero 插件浏览器插件Mac 客户端Windows 客户端微信小程序
定价
高级版会员购买积分包购买API积分包
服务
文献检索文档翻译深度研究API 文档MCP 服务
关于我们
关于 Suppr公司介绍联系我们用户协议隐私条款
关注我们

Suppr 超能文献

核心技术专利：CN118964589B侵权必究
粤ICP备2023148730 号-1Suppr @ 2026

文献检索

告别复杂PubMed语法，用中文像聊天一样搜索，搜遍4000万医学文献。AI智能推荐，让科研检索更轻松。

立即免费搜索

文件翻译

保留排版，准确专业，支持PDF/Word/PPT等文件格式，支持 12+语言互译。

免费翻译文档

深度研究

AI帮你快速写综述，25分钟生成高质量综述，智能提取关键信息，辅助科研写作。

立即免费体验

面部深度伪造检测挑战赛。

The Face Deepfake Detection Challenge.

作者信息

Guarnera Luca, Giudice Oliver, Guarnera Francesco, Ortis Alessandro, Puglisi Giovanni, Paratore Antonino, Bui Linh M Q, Fontani Marco, Coccomini Davide Alessandro, Caldelli Roberto, Falchi Fabrizio, Gennaro Claudio, Messina Nicola, Amato Giuseppe, Perelli Gianpaolo, Concas Sara, Cuccu Carlo, Orrù Giulia, Marcialis Gian Luca, Battiato Sebastiano

机构信息

Department of Mathematics and Computer Science, University of Catania, 95125 Catania, Italy.

Applied Research Team, IT Department, Banca d'Italia, 00044 Rome, Italy.

出版信息

J Imaging. 2022 Sep 28;8(10):263. doi: 10.3390/jimaging8100263.

DOI:10.3390/jimaging8100263
PMID:36286357
原文链接:https://pmc.ncbi.nlm.nih.gov/articles/PMC9605671/
Abstract

Multimedia data manipulation and forgery has never been easier than today, thanks to the power of Artificial Intelligence (AI). AI-generated fake content, commonly called Deepfakes, have been raising new issues and concerns, but also new challenges for the research community. The Deepfake detection task has become widely addressed, but unfortunately, approaches in the literature suffer from generalization issues. In this paper, the Face Deepfake Detection and Reconstruction Challenge is described. Two different tasks were proposed to the participants: (i) creating a Deepfake detector capable of working in an "in the wild" scenario; (ii) creating a method capable of reconstructing original images from Deepfakes. Real images from CelebA and FFHQ and Deepfake images created by StarGAN, StarGAN-v2, StyleGAN, StyleGAN2, AttGAN and GDWCT were collected for the competition. The winning teams were chosen with respect to the highest classification accuracy value (Task I) and "minimum average distance to Manhattan" (Task II). Deep Learning algorithms, particularly those based on the architecture, achieved the best results in Task I. No winners were proclaimed for Task II. A detailed discussion of teams' proposed methods with corresponding ranking is presented in this paper.

摘要

得益于人工智能（AI）的强大力量，多媒体数据处理与伪造如今变得前所未有的容易。人工智能生成的虚假内容，通常称为深度伪造，引发了新的问题和担忧，但也给研究界带来了新的挑战。深度伪造检测任务已得到广泛关注，但遗憾的是，文献中的方法存在泛化问题。本文描述了面部深度伪造检测与重建挑战赛。向参与者提出了两项不同的任务：（i）创建一个能够在“真实场景”中工作的深度伪造检测器；（ii）创建一种能够从深度伪造图像中重建原始图像的方法。比赛收集了来自CelebA和FFHQ的真实图像以及由StarGAN、StarGAN-v2、StyleGAN、StyleGAN2、AttGAN和GDWCT创建的深度伪造图像。根据最高分类准确率值（任务一）和“到曼哈顿的最小平均距离”（任务二）选出获胜团队。深度学习算法，特别是基于该架构的算法，在任务一中取得了最佳成绩。任务二未宣布获胜者。本文详细讨论了各团队提出的方法及相应排名。

https://cdn.ncbi.nlm.nih.gov/pmc/blobs/995e/9605671/0fe4dc4be3ef/jimaging-08-00263-g011.jpg
https://cdn.ncbi.nlm.nih.gov/pmc/blobs/995e/9605671/4a1d2427defc/jimaging-08-00263-g001.jpg
https://cdn.ncbi.nlm.nih.gov/pmc/blobs/995e/9605671/973b341399d3/jimaging-08-00263-g002.jpg
https://cdn.ncbi.nlm.nih.gov/pmc/blobs/995e/9605671/8aafb78cca56/jimaging-08-00263-g003.jpg
https://cdn.ncbi.nlm.nih.gov/pmc/blobs/995e/9605671/176bf9d4d4cf/jimaging-08-00263-g004.jpg
https://cdn.ncbi.nlm.nih.gov/pmc/blobs/995e/9605671/851603d78309/jimaging-08-00263-g005.jpg
https://cdn.ncbi.nlm.nih.gov/pmc/blobs/995e/9605671/204f1cd581fc/jimaging-08-00263-g006.jpg
https://cdn.ncbi.nlm.nih.gov/pmc/blobs/995e/9605671/717a145e7a1c/jimaging-08-00263-g007.jpg
https://cdn.ncbi.nlm.nih.gov/pmc/blobs/995e/9605671/1f82fa81bfaa/jimaging-08-00263-g008.jpg
https://cdn.ncbi.nlm.nih.gov/pmc/blobs/995e/9605671/c59d4d54c62a/jimaging-08-00263-g009.jpg
https://cdn.ncbi.nlm.nih.gov/pmc/blobs/995e/9605671/66c882be6026/jimaging-08-00263-g010.jpg
https://cdn.ncbi.nlm.nih.gov/pmc/blobs/995e/9605671/0fe4dc4be3ef/jimaging-08-00263-g011.jpg
https://cdn.ncbi.nlm.nih.gov/pmc/blobs/995e/9605671/4a1d2427defc/jimaging-08-00263-g001.jpg
https://cdn.ncbi.nlm.nih.gov/pmc/blobs/995e/9605671/973b341399d3/jimaging-08-00263-g002.jpg
https://cdn.ncbi.nlm.nih.gov/pmc/blobs/995e/9605671/8aafb78cca56/jimaging-08-00263-g003.jpg
https://cdn.ncbi.nlm.nih.gov/pmc/blobs/995e/9605671/176bf9d4d4cf/jimaging-08-00263-g004.jpg
https://cdn.ncbi.nlm.nih.gov/pmc/blobs/995e/9605671/851603d78309/jimaging-08-00263-g005.jpg
https://cdn.ncbi.nlm.nih.gov/pmc/blobs/995e/9605671/204f1cd581fc/jimaging-08-00263-g006.jpg
https://cdn.ncbi.nlm.nih.gov/pmc/blobs/995e/9605671/717a145e7a1c/jimaging-08-00263-g007.jpg
https://cdn.ncbi.nlm.nih.gov/pmc/blobs/995e/9605671/1f82fa81bfaa/jimaging-08-00263-g008.jpg
https://cdn.ncbi.nlm.nih.gov/pmc/blobs/995e/9605671/c59d4d54c62a/jimaging-08-00263-g009.jpg
https://cdn.ncbi.nlm.nih.gov/pmc/blobs/995e/9605671/66c882be6026/jimaging-08-00263-g010.jpg
https://cdn.ncbi.nlm.nih.gov/pmc/blobs/995e/9605671/0fe4dc4be3ef/jimaging-08-00263-g011.jpg

相似文献

1
The Face Deepfake Detection Challenge.面部深度伪造检测挑战赛。
J Imaging. 2022 Sep 28;8(10):263. doi: 10.3390/jimaging8100263.
2
A Review of Image Processing Techniques for Deepfakes.深度伪造的图像处理技术综述。
Sensors (Basel). 2022 Jun 16;22(12):4556. doi: 10.3390/s22124556.
3
Deepfakes Generation and Detection: A Short Survey.深度伪造的生成与检测：简要综述
J Imaging. 2023 Jan 13;9(1):18. doi: 10.3390/jimaging9010018.
4
Audio deepfakes: A survey.音频深度伪造：一项调查。
Front Big Data. 2023 Jan 9;5:1001063. doi: 10.3389/fdata.2022.1001063. eCollection 2022.
5
Comparative Analysis of Deepfake Image Detection Method Using Convolutional Neural Network.基于卷积神经网络的深度伪造图像检测方法的比较分析。
Comput Intell Neurosci. 2021 Dec 16;2021:3111676. doi: 10.1155/2021/3111676. eCollection 2021.
6
An efficient deepfake video detection using robust deep learning.使用稳健深度学习的高效深度伪造视频检测
Heliyon. 2024 Feb 9;10(5):e25757. doi: 10.1016/j.heliyon.2024.e25757. eCollection 2024 Mar 15.
7
On the Generalization of Deep Learning Models in Video Deepfake Detection.关于深度学习模型在视频深度伪造检测中的泛化
J Imaging. 2023 Apr 29;9(5):89. doi: 10.3390/jimaging9050089.
8
Deepfake forensics analysis: An explainable hierarchical ensemble of weakly supervised models.深度伪造取证分析：一种弱监督模型的可解释分层集成
Forensic Sci Int Synerg. 2022 Jan 27;4:100217. doi: 10.1016/j.fsisyn.2022.100217. eCollection 2022.
9
Depth map guided triplet network for deepfake face detection.用于深度伪造人脸检测的深度图引导三元组网络。
Neural Netw. 2023 Feb;159:34-42. doi: 10.1016/j.neunet.2022.11.031. Epub 2022 Dec 5.
10
Deepfake attack prevention using steganography GANs.使用隐写术生成对抗网络防止深度伪造攻击
PeerJ Comput Sci. 2022 Oct 20;8:e1125. doi: 10.7717/peerj-cs.1125. eCollection 2022.

引用本文的文献

1
Deepfake Media Forensics: Status and Future Challenges.深度伪造媒体取证：现状与未来挑战。
J Imaging. 2025 Feb 28;11(3):73. doi: 10.3390/jimaging11030073.
2
Possible Health Benefits and Risks of DeepFake Videos: A Qualitative Study in Nursing Students.深度伪造视频可能带来的健康益处与风险：一项针对护理专业学生的定性研究
Nurs Rep. 2024 Oct 3;14(4):2746-2757. doi: 10.3390/nursrep14040203.
3
Detecting images generated by diffusers.检测扩散器生成的图像。

本文引用的文献

1
Fighting Deepfakes by Detecting GAN DCT Anomalies.通过检测生成对抗网络离散余弦变换异常来对抗深度伪造
J Imaging. 2021 Jul 30;7(8):128. doi: 10.3390/jimaging7080128.
2
AttGAN: Facial Attribute Editing by Only Changing What You Want.AttGAN：仅通过改变你想要改变的内容来进行面部属性编辑。
IEEE Trans Image Process. 2019 Nov;28(11):5464-5478. doi: 10.1109/TIP.2019.2916751. Epub 2019 May 20.
3
A mathematical analysis of the DCT coefficient distributions for images.图像离散余弦变换（DCT）系数分布的数学分析
PeerJ Comput Sci. 2024 Jul 10;10:e2127. doi: 10.7717/peerj-cs.2127. eCollection 2024.
4
On the Generalization of Deep Learning Models in Video Deepfake Detection.关于深度学习模型在视频深度伪造检测中的泛化
J Imaging. 2023 Apr 29;9(5):89. doi: 10.3390/jimaging9050089.
IEEE Trans Image Process. 2000;9(10):1661-6. doi: 10.1109/83.869177.