Filters








319 Hits in 7.8 sec

Real-centric Consistency Learning for Deepfake Detection [article]

Ruiqi Zha, Zhichao Lian, Qianmu Li, Siqi Gu
2022 arXiv   pre-print
Besides, a hard negative fusion method is designed to improve the discrimination of negative marginal features with the help of supervised contrastive margin loss we developed.  ...  Essentially, the target of deepfake detection problem is to represent natural faces and fake faces at the representation space discriminatively, and it reminds us whether we could optimize the feature  ...  , we locate the deepfake detection problem on learning the invariant representations for natural faces and forgery faces both.  ... 
arXiv:2205.07201v1 fatcat:mtpr3z566jef7e5xb2d3aok3py

Towards Intrinsic Common Discriminative Features Learning for Face Forgery Detection using Adversarial Learning [article]

Wanyi Zhuang, Qi Chu, Haojie Yuan, Changtao Miao, Bin Liu, Nenghai Yu
2022 arXiv   pre-print
to learn intrinsic common discriminative features for face forgery detection.  ...  Existing face forgery detection methods usually treat face forgery detection as a binary classification problem and adopt deep convolution neural networks to learn discriminative features.  ...  All the detected faces are cropped (enlarged by a factor of 1.3) around the center of the face, and then resized to 299 × 299.  ... 
arXiv:2207.03776v1 fatcat:flyjj6saczcall6b6vmn4vsp3e

Multi-attentional Deepfake Detection [article]

Hanqing Zhao, Wenbo Zhou, Dongdong Chen, Tianyi Wei, Weiming Zhang, Nenghai Yu
2021 arXiv   pre-print
Face forgery by deepfake is widely spread over the internet and has raised severe societal concerns.  ...  in shallow features; 3) aggregate the low-level textural feature and high-level semantic features guided by the attention maps.  ...  [32] uses frequency-aware decomposition and local frequency statistic to expose deepfake artifacts in frequency domain and achieves state-of-the-art performance.  ... 
arXiv:2103.02406v3 fatcat:syk6womid5hhtl4c4udwwskpny

Feature Focus: Towards Explainable and Transparent Deep Face Morphing Attack Detectors

Clemens Seibold, Anna Hilsmann, Peter Eisert
2021 Computers  
genuine face images or considered for an individual morphed face image.  ...  In this paper, we present Feature Focus, a new transparent face morphing detector based on a modified VGG-A architecture and an additional feature shaping loss function, as well as Focused Layer-wise Relevance  ...  areas (partial morphs) • Analysis of the features' discrimination power learned by DNNs for face morphing attack detection and its relation to interpretability via FLRP and LRP • Reliable and accurate  ... 
doi:10.3390/computers10090117 fatcat:s3rsu3rmzjaybhiuu3o6hvkwz4

Making DeepFakes more spurious: evading deep face forgery detection via trace removal attack [article]

Chi Liu, Huajie Chen, Tianqing Zhu, Jun Zhang, Wanlei Zhou
2022 arXiv   pre-print
Each discriminator is responsible for one individual trace representation to avoid cross-trace interference.  ...  Then a trace removal network (TR-Net) is proposed based on an adversarial learning framework involving one generator and multiple discriminators.  ...  It involves a two-stream collaborative network that combines frequency-aware decomposition and local frequency statistics to learn frequency-aware clues.  ... 
arXiv:2203.11433v1 fatcat:oxidlgwn7rdnxcnt7ab2jl5mzm

ID-Reveal: Identity-aware DeepFake Video Detection [article]

Davide Cozzolino and Andreas Rössler and Justus Thies and Matthias Nießner and Luisa Verdoliva
2021 arXiv   pre-print
To this end, we introduce ID-Reveal, a new approach that learns temporal facial features, specific of how a person moves while talking, by means of metric learning coupled with an adversarial training  ...  A major challenge in DeepFake forgery detection is that state-of-the-art algorithms are mostly trained to detect a specific fake method.  ...  Government is authorized to reproduce and distribute reprints for Governmental purposes notwithstanding any copyright notation thereon.  ... 
arXiv:2012.02512v3 fatcat:j7do67ktenbstoye66xucybfsy

A Full-Image Full-Resolution End-to-End-Trainable CNN Framework for Image Forgery Detection

Francesco Marra, Diego Gragnaniello, Luisa Verdoliva, Giovanni Poggi
2020 IEEE Access  
This is not a problem for high-level vision problems, where discriminative features are barely affected by resizing.  ...  For more information, see https://creativecommons.org/licenses/by/4.0/ VOLUME 8, 2020  ...  This is very likely due to the loss of precious high-frequency details induced by resizing.  ... 
doi:10.1109/access.2020.3009877 fatcat:kvqs5iu7svcklmado3x3z7bony

A Full-Image Full-Resolution End-to-End-Trainable CNN Framework for Image Forgery Detection [article]

Francesco Marra, Diego Gragnaniello, Luisa Verdoliva, Giovanni Poggi
2019 arXiv   pre-print
This is not a problem for high-level vision problems, where discriminative features are barely affected by resizing.  ...  In this work, we propose a CNN-based image forgery detection framework which makes decisions based on full-resolution information gathered from the whole image.  ...  The loss back-propagates through the net up to to individual patches, allowing the feature extractor to learn which information is the most discriminative for the final decision, and adapting the classifier  ... 
arXiv:1909.06751v1 fatcat:vwh75pn6ofac5dcf3oy742team

2021 Index IEEE Transactions on Image Processing Vol. 30

2021 IEEE Transactions on Image Processing  
The Author Index contains the primary entry for each item, listed under the first author's name.  ...  The primary entry includes the coauthors' names, the title of the paper or other item, and its location, specified by the publication abbreviation, year, month, and inclusive pagination.  ...  ., +, TIP 2021 5793-5806 SFace: Sigmoid-Constrained Hypersphere Loss for Robust Face Recognition. Zhong, Y., +, TIP 2021 2587-2598 Structure-Coherent Deep Feature Learning for Robust Face Alignment.  ... 
doi:10.1109/tip.2022.3142569 fatcat:z26yhwuecbgrnb2czhwjlf73qu

2021 Index IEEE Transactions on Multimedia Vol. 23

2021 IEEE transactions on multimedia  
The Author Index contains the primary entry for each item, listed under the first author's name.  ...  The primary entry includes the coauthors' names, the title of the paper or other item, and its location, specified by the publication abbreviation, year, month, and inclusive pagination.  ...  ., +, TMM 2021 4483-4490 Learning (artificial intelligence) 3D Face Reconstruction From A Single Image Assisted by 2D Face Images in the Wild.  ... 
doi:10.1109/tmm.2022.3141947 fatcat:lil2nf3vd5ehbfgtslulu7y3lq

Watch your Up-Convolution: CNN Based Generative Deep Neural Networks are Failing to Reproduce Spectral Distributions [article]

Ricard Durall and Margret Keuper and Janis Keuper
2020 arXiv   pre-print
We show that this approach not only allows to train spectral consistent GANs that are avoiding high frequency errors.  ...  This effect is independent of the underlying architecture and we show that it can be used to easily detect generated data like deepfakes with up to 100% accuracy on public benchmarks.  ...  As expected based on our theoretical analysis in sec. 2.3, the observed effects can not be corrected by a single, learned 3 × 3 filter, even for large values λ.  ... 
arXiv:2003.01826v1 fatcat:o6rj4eo4bjfx7jqbn75d4qpbdu

Watch Your Up-Convolution: CNN Based Generative Deep Neural Networks Are Failing to Reproduce Spectral Distributions

Ricard Durall, Margret Keuper, Janis Keuper
2020 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)  
We show that this approach not only allows to train spectral consistent GANs that are avoiding high frequency errors.  ...  This effect is independent of the underlying architecture and we show that it can be used to easily detect generated data like deepfakes with up to 100% accuracy on public benchmarks.  ...  As expected based on our theoretical analysis in sec. 2.3, the observed effects can not be corrected by a single, learned 3 × 3 filter, even for large values λ.  ... 
doi:10.1109/cvpr42600.2020.00791 dblp:conf/cvpr/DurallKK20 fatcat:fba4fwrwf5hcpf2jm33fdhuexq

A Review of Deep Learning-based Approaches for Deepfake Content Detection [article]

Leandro A. Passos, Danilo Jodas, Kelton A. P. da Costa, Luis A. Souza Júnior, Danilo Colombo, João Paulo Papa
2022 arXiv   pre-print
This paper presents a comprehensive review of recent studies for deepfake content detection using deep learning-based approaches.  ...  We aim to broaden the state-of-the-art research by systematically reviewing the different categories of fake content detection.  ...  [78] showed promising results for counterfeit fingerprint detection using a single deep learning architecture.  ... 
arXiv:2202.06095v1 fatcat:27ogp2kj4jayvmoe5xbhhi33li

Delving into the Local: Dynamic Inconsistency Learning for DeepFake Video Detection

Zhihao Gu, Yang Chen, Taiping Yao, Shouhong Ding, Jilin Li, Lizhuang Ma
2022 PROCEEDINGS OF THE THIRTIETH AAAI CONFERENCE ON ARTIFICIAL INTELLIGENCE AND THE TWENTY-EIGHTH INNOVATIVE APPLICATIONS OF ARTIFICIAL INTELLIGENCE CONFERENCE  
Existing deepfake video detection approaches attempt to capture the discrim- inative features between real and fake faces based on tem- poral modelling.  ...  an efficient indica- tor for DeepFake video detection.  ...  Therefore, it is of great importance to develop effective methods for face forgery detection. Recently, significant progress has been achieved in Deep-Fake detection.  ... 
doi:10.1609/aaai.v36i1.19955 fatcat:y7ukh3qlirhj7mbq4643e23viy

A Comprehensive Review of Deep-Learning-Based Methods for Image Forensics

Ivan Castillo Castillo Camacho, Kai Wang
2021 Journal of Imaging  
Furthermore, image forgeries are presently so realistic that it becomes difficult for the naked eye to differentiate between fake and real media.  ...  With this review it can be observed that even if image forgeries are becoming easy to create, there are several options to detect each kind of them.  ...  In [68] authors proposed a 9-layer CNN that is directly fed with 64 × 64 image pixel values with no special features, making the discriminative features self-learned by the network.  ... 
doi:10.3390/jimaging7040069 pmid:34460519 pmcid:PMC8321383 fatcat:72zd7nyaifhvlgcxv22zztpm4y
« Previous Showing results 1 — 15 out of 319 results