Show simple record

dc.contributor.authorWang, Changjin
dc.contributor.authorSharifzadeh, Hamid
dc.contributor.authorVarastehpour, Soheil
dc.contributor.authorArdekani, Iman
dc.date.accessioned2023-11-16T00:25:22Z
dc.date.available2023-11-16T00:25:22Z
dc.date.issued2023-08-21
dc.identifier.urihttps://hdl.handle.net/10652/6152
dc.description.abstractThe rise of generative artificial intelligence (GenAI) has made it increasingly possible to use Deepfakes technology to generate fake pictures and videos. While this technology has benefits, it also has downsides such as spreading misinformation and endangering public interests. To address this issue, researchers have proposed various deep forgery detection algorithms and have achieved remarkable results. However, a common problem regarding these detection methods is that while in-library detection can usually achieve high accuracy, their performance is significantly degraded in cross-library detection. This indicates a severe problem of insufficient generalisation ability. To better compare the performance differences between various detection methods, this paper analyses the detection performance of the six established models of Two-stream, MesoNet, HeadPose, FWA, VA, and Multi-task. To ensure consistency, we employ a uniform evaluation framework as a benchmark for comparison. We conduct extensive intra-library and crosslibrary tests to evaluate these methods’ generalisation ability by utilising accuracy and error rate as key evaluation criteria for our experiments. Additionally, we further explore areas for improvement by analysing the impact of data augmentation, dataset partitioning, and threshold selection on the performance of these detection methods. Our comparative experiments are conducted on three existing fake face video datasets, including FaceForensics++, DeepfakeTIMIT, and Celeb-DF. Our research findings indicate the database partitioning method has a direct impact on the detector’s performance, and to enhance generalisation performance, the database should be divided person-based manually. The effectiveness of data augmentation techniques in improving cross-library performance is generally limited, and setting the threshold directly using source domain data often leads to a high error rate in the target domain. The findings of this paper provide insights into the development of more effective detection methods to combat the harmful effectsen_NZ
dc.language.isoenen_NZ
dc.relation.urihttps://pstnet.ca/callforpapers.htmlen_NZ
dc.rightsAll rights reserveden_NZ
dc.subjectdeepfakesen_NZ
dc.subjectdetectionen_NZ
dc.subjectcross-library generalisationen_NZ
dc.subjecttransfer learningen_NZ
dc.titleAnalysis and comparison of deepfakes detection methods for cross-library generalisationen_NZ
dc.typeConference Contribution - Paper in Published Proceedingsen_NZ
dc.date.updated2023-11-09T13:30:30Z
dc.rights.holderAuthorsen_NZ
dc.subject.marsden4602 Artificial intelligenceen_NZ
dc.identifier.bibliographicCitationWang, C., Sharifzadeh, H.., Varastehpour, S., & Ardekani, I. (2023, August 21-23). Analysis and comparison of deepfakes detection methods for cross-library generalisation [Paper presentation] 20th Annual International Conference on Privacy, Security & Trust, Copenhagen, Denmark (PST2023) (pp. 1-6). https://hdl.handle.net/10652/6152en_NZ
unitec.publication.spage1en_NZ
unitec.publication.lpage6en_NZ
unitec.conference.titleAnnual International Conference on Privacy, Security & Trust, Copenhagen, Denmark (PST2023)en_NZ
unitec.conference.locationCopenhagen, Demarken_NZ
unitec.conference.sdate2023-08-21
unitec.conference.edate2023-08-23
unitec.peerreviewedyesen_NZ
dc.contributor.affiliationUnitec, Te Pūkengaen_NZ
dc.contributor.affiliationTe Pūkengaen_NZ
unitec.identifier.roms71060en_NZ
unitec.institution.studyareaComputingen_NZ


Files in this item

Thumbnail

This item appears in

Show simple record