Detecting Compressed Deepfake Videos in Social Networks Using Frame-Temporality Two-Stream Convolutional Network

doi:10.1109/TCSVT.2021.3074259

CORC > 自动化研究所 > 中国科学院自动化研究所 > 智能感知与计算研究中心

	Detecting Compressed Deepfake Videos in Social Networks Using Frame-Temporality Two-Stream Convolutional Network
	Hu, Juan 3; Liao, Xin 2,3; Wang, Wei 1; Qin, Zheng 3
刊名	IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY
	2022-03-01
卷号	32 期号:3 页码:1089-1102
关键词	Videos Information integrity Feature extraction Streaming media Faces Forensics Social networking (online) Video forensics compressed Deepfake videos frame-level stream temporality-level stream
ISSN号	1051-8215
DOI	10.1109/TCSVT.2021.3074259
通讯作者	Liao, Xin(xinliao@hnu.edu.cn)
英文摘要	The development of technologies that can generate Deepfake videos is expanding rapidly. These videos are easily synthesized without leaving obvious traces of manipulation. Though forensically detection in high-definition video datasets has achieved remarkable results, the forensics of compressed videos is worth further exploring. In fact, compressed videos are common in social networks, such as videos from Instagram, Wechat, and Tiktok. Therefore, how to identify compressed Deepfake videos becomes a fundamental issue. In this paper, we propose a two-stream method by analyzing the frame-level and temporality-level of compressed Deepfake videos. Since the video compression brings lots of redundant information to frames, the proposed frame-level stream gradually prunes the network to prevent the model from fitting the compression noise. Aiming at the problem that the temporal consistency in Deepfake videos might be ignored, we apply a temporality-level stream to extract temporal correlation features. When combined with scores from the two streams, our proposed method performs better than the state-of-the-art methods in compressed Deepfake videos detection.
资助项目	National Natural Science Foundation of China[61972142] ; National Natural Science Foundation of China[61972395] ; National Natural Science Foundation of China[61772191] ; Hunan Provincial Natural Science Foundation of China[2020JJ4212] ; Key Lab of Information Network Security, Ministry of Public Security[C20611] ; Science and Technology Program of Changsha[kq2004021]
WOS关键词	FORENSICS
WOS研究方向	Engineering
语种	英语
出版者	IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
WOS记录号	WOS:000766700400017
资助机构	National Natural Science Foundation of China ; Hunan Provincial Natural Science Foundation of China ; Key Lab of Information Network Security, Ministry of Public Security ; Science and Technology Program of Changsha
内容类型	期刊论文
源URL	[http://ir.ia.ac.cn/handle/173211/48087]
专题	自动化研究所_智能感知与计算研究中心
通讯作者	Liao, Xin
作者单位	1.Chinese Acad Sci, Inst Automat, Natl Lab Pattern Recognit, Beijing 100190, Peoples R China 2.Chinese Acad Sci, Inst Informat Engn, State Key Lab Informat Secur, Beijing 100093, Peoples R China 3.Hunan Univ, Coll Comp Sci & Elect Engn, Changsha 410082, Peoples R China
推荐引用方式 GB/T 7714	Hu, Juan,Liao, Xin,Wang, Wei,et al. Detecting Compressed Deepfake Videos in Social Networks Using Frame-Temporality Two-Stream Convolutional Network[J]. IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY,2022,32(3):1089-1102.
APA	Hu, Juan,Liao, Xin,Wang, Wei,&Qin, Zheng.(2022).Detecting Compressed Deepfake Videos in Social Networks Using Frame-Temporality Two-Stream Convolutional Network.IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY,32(3),1089-1102.
MLA	Hu, Juan,et al."Detecting Compressed Deepfake Videos in Social Networks Using Frame-Temporality Two-Stream Convolutional Network".IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY 32.3(2022):1089-1102.