WWW: Where, Which and Whatever Enhancing Interpretability in Multimodal Deepfake Detection

All current benchmarks for multimodal deepfake detection manipulate entire frames using various generation techniques, resulting in oversaturated detection accuracies exceeding 94% at the video-level classification. However, these benchmarks struggle to detect dynamic deepfake attacks with challengi...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Jung, Juho, Lee, Sangyoun, Kang, Jooeon, Na, Yunjin
Format: Artikel
Sprache:eng
Schlagworte:
Online-Zugang:Volltext bestellen
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
container_end_page
container_issue
container_start_page
container_title
container_volume
creator Jung, Juho
Lee, Sangyoun
Kang, Jooeon
Na, Yunjin
description All current benchmarks for multimodal deepfake detection manipulate entire frames using various generation techniques, resulting in oversaturated detection accuracies exceeding 94% at the video-level classification. However, these benchmarks struggle to detect dynamic deepfake attacks with challenging frame-by-frame alterations presented in real-world scenarios. To address this limitation, we introduce FakeMix, a novel clip-level evaluation benchmark aimed at identifying manipulated segments within both video and audio, providing insight into the origins of deepfakes. Furthermore, we propose novel evaluation metrics, Temporal Accuracy (TA) and Frame-wise Discrimination Metric (FDM), to assess the robustness of deepfake detection models. Evaluating state-of-the-art models against diverse deepfake benchmarks, particularly FakeMix, demonstrates the effectiveness of our approach comprehensively. Specifically, while achieving an Average Precision (AP) of 94.2% at the video-level, the evaluation of the existing models at the clip-level using the proposed metrics, TA and FDM, yielded sharp declines in accuracy to 53.1%, and 52.1%, respectively.
doi_str_mv 10.48550/arxiv.2408.02954
format Article
fullrecord <record><control><sourceid>arxiv_GOX</sourceid><recordid>TN_cdi_arxiv_primary_2408_02954</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>2408_02954</sourcerecordid><originalsourceid>FETCH-arxiv_primary_2408_029543</originalsourceid><addsrcrecordid>eNqFjjEOgkAQAK-xMOoDrLwHKJ4ICdoqRgs7k0tsyAqLbDwOcq5Efi8Se6uZYooRYrpSXhCFoVqCe1Pj-YGKPOVvwmAorlrrrdQFOpx3oLSQYLPOgLFBJ2NbgE3J3uXJMrraIcONDHErycrzyzCVVQZG7hHrHB7YCWPKVNmxGORgnjj5cSRmh_iyOy76i6R2VIJrk-9N0t-s_xcfT-E_3w</addsrcrecordid><sourcetype>Open Access Repository</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype></control><display><type>article</type><title>WWW: Where, Which and Whatever Enhancing Interpretability in Multimodal Deepfake Detection</title><source>arXiv.org</source><creator>Jung, Juho ; Lee, Sangyoun ; Kang, Jooeon ; Na, Yunjin</creator><creatorcontrib>Jung, Juho ; Lee, Sangyoun ; Kang, Jooeon ; Na, Yunjin</creatorcontrib><description>All current benchmarks for multimodal deepfake detection manipulate entire frames using various generation techniques, resulting in oversaturated detection accuracies exceeding 94% at the video-level classification. However, these benchmarks struggle to detect dynamic deepfake attacks with challenging frame-by-frame alterations presented in real-world scenarios. To address this limitation, we introduce FakeMix, a novel clip-level evaluation benchmark aimed at identifying manipulated segments within both video and audio, providing insight into the origins of deepfakes. Furthermore, we propose novel evaluation metrics, Temporal Accuracy (TA) and Frame-wise Discrimination Metric (FDM), to assess the robustness of deepfake detection models. Evaluating state-of-the-art models against diverse deepfake benchmarks, particularly FakeMix, demonstrates the effectiveness of our approach comprehensively. Specifically, while achieving an Average Precision (AP) of 94.2% at the video-level, the evaluation of the existing models at the clip-level using the proposed metrics, TA and FDM, yielded sharp declines in accuracy to 53.1%, and 52.1%, respectively.</description><identifier>DOI: 10.48550/arxiv.2408.02954</identifier><language>eng</language><subject>Computer Science - Computer Vision and Pattern Recognition</subject><creationdate>2024-08</creationdate><rights>http://arxiv.org/licenses/nonexclusive-distrib/1.0</rights><oa>free_for_read</oa><woscitedreferencessubscribed>false</woscitedreferencessubscribed></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><link.rule.ids>228,230,780,885</link.rule.ids><linktorsrc>$$Uhttps://arxiv.org/abs/2408.02954$$EView_record_in_Cornell_University$$FView_record_in_$$GCornell_University$$Hfree_for_read</linktorsrc><backlink>$$Uhttps://doi.org/10.48550/arXiv.2408.02954$$DView paper in arXiv$$Hfree_for_read</backlink></links><search><creatorcontrib>Jung, Juho</creatorcontrib><creatorcontrib>Lee, Sangyoun</creatorcontrib><creatorcontrib>Kang, Jooeon</creatorcontrib><creatorcontrib>Na, Yunjin</creatorcontrib><title>WWW: Where, Which and Whatever Enhancing Interpretability in Multimodal Deepfake Detection</title><description>All current benchmarks for multimodal deepfake detection manipulate entire frames using various generation techniques, resulting in oversaturated detection accuracies exceeding 94% at the video-level classification. However, these benchmarks struggle to detect dynamic deepfake attacks with challenging frame-by-frame alterations presented in real-world scenarios. To address this limitation, we introduce FakeMix, a novel clip-level evaluation benchmark aimed at identifying manipulated segments within both video and audio, providing insight into the origins of deepfakes. Furthermore, we propose novel evaluation metrics, Temporal Accuracy (TA) and Frame-wise Discrimination Metric (FDM), to assess the robustness of deepfake detection models. Evaluating state-of-the-art models against diverse deepfake benchmarks, particularly FakeMix, demonstrates the effectiveness of our approach comprehensively. Specifically, while achieving an Average Precision (AP) of 94.2% at the video-level, the evaluation of the existing models at the clip-level using the proposed metrics, TA and FDM, yielded sharp declines in accuracy to 53.1%, and 52.1%, respectively.</description><subject>Computer Science - Computer Vision and Pattern Recognition</subject><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2024</creationdate><recordtype>article</recordtype><sourceid>GOX</sourceid><recordid>eNqFjjEOgkAQAK-xMOoDrLwHKJ4ICdoqRgs7k0tsyAqLbDwOcq5Efi8Se6uZYooRYrpSXhCFoVqCe1Pj-YGKPOVvwmAorlrrrdQFOpx3oLSQYLPOgLFBJ2NbgE3J3uXJMrraIcONDHErycrzyzCVVQZG7hHrHB7YCWPKVNmxGORgnjj5cSRmh_iyOy76i6R2VIJrk-9N0t-s_xcfT-E_3w</recordid><startdate>20240806</startdate><enddate>20240806</enddate><creator>Jung, Juho</creator><creator>Lee, Sangyoun</creator><creator>Kang, Jooeon</creator><creator>Na, Yunjin</creator><scope>AKY</scope><scope>GOX</scope></search><sort><creationdate>20240806</creationdate><title>WWW: Where, Which and Whatever Enhancing Interpretability in Multimodal Deepfake Detection</title><author>Jung, Juho ; Lee, Sangyoun ; Kang, Jooeon ; Na, Yunjin</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-arxiv_primary_2408_029543</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2024</creationdate><topic>Computer Science - Computer Vision and Pattern Recognition</topic><toplevel>online_resources</toplevel><creatorcontrib>Jung, Juho</creatorcontrib><creatorcontrib>Lee, Sangyoun</creatorcontrib><creatorcontrib>Kang, Jooeon</creatorcontrib><creatorcontrib>Na, Yunjin</creatorcontrib><collection>arXiv Computer Science</collection><collection>arXiv.org</collection></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext_linktorsrc</fulltext></delivery><addata><au>Jung, Juho</au><au>Lee, Sangyoun</au><au>Kang, Jooeon</au><au>Na, Yunjin</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>WWW: Where, Which and Whatever Enhancing Interpretability in Multimodal Deepfake Detection</atitle><date>2024-08-06</date><risdate>2024</risdate><abstract>All current benchmarks for multimodal deepfake detection manipulate entire frames using various generation techniques, resulting in oversaturated detection accuracies exceeding 94% at the video-level classification. However, these benchmarks struggle to detect dynamic deepfake attacks with challenging frame-by-frame alterations presented in real-world scenarios. To address this limitation, we introduce FakeMix, a novel clip-level evaluation benchmark aimed at identifying manipulated segments within both video and audio, providing insight into the origins of deepfakes. Furthermore, we propose novel evaluation metrics, Temporal Accuracy (TA) and Frame-wise Discrimination Metric (FDM), to assess the robustness of deepfake detection models. Evaluating state-of-the-art models against diverse deepfake benchmarks, particularly FakeMix, demonstrates the effectiveness of our approach comprehensively. Specifically, while achieving an Average Precision (AP) of 94.2% at the video-level, the evaluation of the existing models at the clip-level using the proposed metrics, TA and FDM, yielded sharp declines in accuracy to 53.1%, and 52.1%, respectively.</abstract><doi>10.48550/arxiv.2408.02954</doi><oa>free_for_read</oa></addata></record>
fulltext fulltext_linktorsrc
identifier DOI: 10.48550/arxiv.2408.02954
ispartof
issn
language eng
recordid cdi_arxiv_primary_2408_02954
source arXiv.org
subjects Computer Science - Computer Vision and Pattern Recognition
title WWW: Where, Which and Whatever Enhancing Interpretability in Multimodal Deepfake Detection
url https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2024-12-23T20%3A28%3A29IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-arxiv_GOX&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=WWW:%20Where,%20Which%20and%20Whatever%20Enhancing%20Interpretability%20in%20Multimodal%20Deepfake%20Detection&rft.au=Jung,%20Juho&rft.date=2024-08-06&rft_id=info:doi/10.48550/arxiv.2408.02954&rft_dat=%3Carxiv_GOX%3E2408_02954%3C/arxiv_GOX%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_id=info:pmid/&rfr_iscdi=true