15 Hits in 1.6 sec

CIAGAN: Conditional Identity Anonymization Generative Adversarial Networks [article]

Maxim Maximov, Ismail Elezi, Laura Leal-Taixé
2020 arXiv   pre-print
We propose and develop CIAGAN, a model for image and video anonymization based on conditional generative adversarial networks.  ...  Unlike previous methods, we have full control over the de-identification (anonymization) procedure, ensuring both anonymization as well as diversity.  ...  Our novel CIAGAN model is based on conditional generative adversarial networks, and faces are anonymized based on a guiding identity signal provided by a siamese network.  ... 
arXiv:2005.09544v1 fatcat:swkfdjrdazbipiarxlgbkdtbbi

CIAGAN: Conditional Identity Anonymization Generative Adversarial Networks

Maxim Maximov, Ismail Elezi, Laura Leal-Taixe
2020 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)  
Figure 1: Given a face image, our network anonymizes the face based on the desired identity. In the figure, it can be seen the variability of the generated faces, controlled from a given label.  ...  In each triplet, the first image is the real image, while the other two images are different anonymized versions of the real image.  ...  Our novel CIAGAN model is based on conditional generative adversarial networks, and faces are anonymized based on a guiding identity signal provided by a siamese network.  ... 
doi:10.1109/cvpr42600.2020.00549 dblp:conf/cvpr/MaximovEL20 fatcat:qy5os5ouozbpvfdlk3dgrkntie

IdentityDP: Differential Private Identification Protection for Face Images [article]

Yunqian Wen, Li Song, Bo Liu, Ming Ding, Rong Xie
2021 arXiv   pre-print
We propose IdentityDP, a face anonymization framework that combines a data-driven deep neural network with a differential privacy (DP) mechanism.  ...  Given a face image, face de-identification, also known as face anonymization, refers to generating another image with similar appearance and the same background, while the real identity is hidden.  ...  Conditioned on face landmark and masked background image of the input image, CIAGAN generates a new fake identity out of the input image to achieve anonymization.  ... 
arXiv:2103.01745v1 fatcat:uev73p3apfbxxlffiih7x3fpzm

Development of a Privacy-Preserving UAV System with Deep Learning-Based Face Anonymization

Harim Lee, Myeung Un Kim, Yeongjun Kim, Hyeonsu Lyu, Hyun Jong Yang
2021 IEEE Access  
Leal-Taixé, "CIAGAN: conditional identity 6 anonymization generative adversarial networks," in 2020 IEEE/CVF 7 ( a post-doctoral researcher in Department of Electrical Engineering, Pohang 28 University  ...  modifier by using generative 70 adversarial networks (GANs).  ... 
doi:10.1109/access.2021.3113186 fatcat:qgvhwuba5ba3fo6y5hw7dw7huu

Privacy preserving human activity recognition framework using an optimized prediction algorithm

Kambala Vijaya Kumar, Jonnadula Harikiran
2022 IAES International Journal of Artificial Intelligence (IJ-AI)  
It anonymizes video content to have adaptive privacy model that defeats attacks from adversaries.  ...  The visual recognition of human actions is made using an underlying adversarial learning process where the anonymization is optimized to have an adaptive privacy model.  ...  [9] proposed a GAN based system known as conditional identity anonymization generative adversarial network (CIAGAN) which supports anonymization and recognition of actions in image and video.  ... 
doi:10.11591/ijai.v11.i1.pp254-264 fatcat:2z7r3ppyq5gyfiz7zxkaecnata

Trusted AI in Multi-agent Systems: An Overview of Privacy and Security for Distributed Learning [article]

Chuan Ma, Jun Li, Kang Wei, Bo Liu, Ming Ding, Long Yuan, Zhu Han, H. Vincent Poor
2022 arXiv   pre-print
The work in [150] has proposed the conditional identity anonymization generative adversarial networks (CIAGAN) model, which can remove the identifying characteristics of faces and bodies while producing  ...  A multi-objective loss function [149] Motion data involving an information-theoretic Concealing user's private identity approach [150] Image and video Conditional generative adversarial networks Removing  ... 
arXiv:2202.09027v2 fatcat:hlu7bopcjrc6zjn2pct57utufy

Perceptual Indistinguishability-Net (PI-Net): Facial Image Obfuscation with Manipulable Semantics [article]

Jia-Wei Chen, Li-Ju Chen, Chia-Mu Yu, Chun-Shien Lu
2021 arXiv   pre-print
DeepPrivacy [21] and CIAGAN [30] are conditional GANs (CGANs), generating anonymized images.  ...  Through the adversarial learning that uses the generator as an anonymizer to modify sensitive information and the discriminator as a facial identifier, the trained generator can synthesize high-quality  ... 
arXiv:2104.01753v2 fatcat:vdbhpaj7sje2plfdftqi7rggsu

The UU-Net: Reversible Face De-Identification for Visual Surveillance Video Footage [article]

Hugo Proença
2020 arXiv   pre-print
The proposed solution is landmarks-free and uses a conditional generative adversarial network to generate synthetic faces that preserve pose, lighting, background information and even facial expressions  ...  Our solution is able to generate a photo realistic de-identified stream that meets the data protection regulations and can be publicly released under minimal privacy constraints.  ...  We designed a two-stage learning process, with a conditional generative adversarial network composed of two different entities (an encoder and a decoder) that have as common goal to fool an adversarial  ... 
arXiv:2007.04316v1 fatcat:lzjaitgezrd6zfqnlphb35bdty

Privacy–Enhancing Face Biometrics: A Comprehensive Survey

Blaz Meden, Peter Rot, Philipp Terhorst, Naser Damer, Arjan Kuijper, Walter J. Scheirer, Arun Ross, Peter Peer, Vitomir Struc
2021 IEEE Transactions on Information Forensics and Security  
Conditional GANs were also used by Maximov et al. for their Conditional Identity Anonymization GAN model (CIAGAN) [228] .  ...  CIAGAN, in essence, implements a face swapping procedure using an encoder-decoder type generator network trained in an adversarial manner.  ... 
doi:10.1109/tifs.2021.3096024 fatcat:z5kvij6g7vgx3b24narxdyp2py

Invertible Mask Network for Face Privacy-Preserving [article]

Yang Yang, Yiyang Huang, Ming Shi, Kejiang Chen, Weiming Zhang, Nenghai Yu
2022 arXiv   pre-print
In IMN, we introduce a Mask-net to generate "Mask" face firstly.  ...  To achieve the naturalness of the processed face and the recoverability of the original protected face, this paper proposes face privacy-preserving method based on Invertible "Mask" Network (IMN).  ...  proposed CIAGAN [4] , a model for image and video anonymization based on conditional generative adversarial networks.  ... 
arXiv:2204.08895v1 fatcat:v7fkgelfgvhk7jvyahrcjwhnfm

Unsupervised Audiovisual Synthesis via Exemplar Autoencoders [article]

Kangle Deng and Aayush Bansal and Deva Ramanan
2021 arXiv   pre-print
Ciagan: Conditional identity anonymization generative adversarial networks. In CVPR, 2020. Sachit Menon, Alexandru Damian, Shijia Hu, Nikhil Ravi, and Cynthia Rudin.  ...  Recent approaches for image anonymization make use of generative models that "deidentify" data without degradive blurring by retargeting each face to a generic identity (e.g., make everyone in a dataset  ... 
arXiv:2001.04463v3 fatcat:ef7dbok5bjhn3or4bj5d45rtre

Controllable Data Generation by Deep Learning: A Review [article]

Shiyu Wang, Yuanqi Du, Xiaojie Guo, Bo Pan, Liang Zhao
2022 arXiv   pre-print
This article provides a systematic review of this promising research area, commonly known as controllable deep data generation.  ...  Finally, the promising future directions of controllable deep data generation are highlighted and five potential challenges are identified.  ...  CIAGAN anonymizes the face of the image based on the identity that is encoded from a reference data via CNN [198] . In this case, the objective function is aligned with Eq. 22.  ... 
arXiv:2207.09542v2 fatcat:ey6v72rkxjbghdw63y2v2kjcde

MARVEL - D5.2: Technical evaluation and progress against benchmarks – initial version

Toni Heittola, Tuomas Virtanen
2022 Zenodo  
Advances in Generative Adversarial Networks (GAN) have allowed proposing different GAN-based video anonymisation solutions by swapping the original face with natural-looking faces of another identity with  ...  In general, with the provided models, most of the methods can generate natural and pose-preserving faces of the specified identity with the high-quality face images, apart from CIAGAN which tends to produce  ...  KPI-O1-E3-1: Number of incorporated safety mechanisms (e.g. for privacy, voice anonymization) ≥ 3.  ... 
doi:10.5281/zenodo.6322699 fatcat:d5lpwby5szg4fih77sp5rjoyae

Table of Contents

2020 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)  
: Conditional Identity Anonymization Generative Adversarial Networks 5446 Maxim Maximov (Technical University Munich), Ismail Elezi (University of Venice), and Laura Leal-Taixé (Technical University  ...  Kenyon (Los Alamos National Laboratory, Los Alamos, NM) DOA-GAN: Dual-Order Attentive Generative Adversarial Network for Image Copy-Move Forgery Detection and Localization 4675 Ashraful Islam (Rensselaer  ... 
doi:10.1109/cvpr42600.2020.00004 fatcat:c7els2kee5cq7lh6cemeqhdcoa

Exploiting Contextual Information with Deep Neural Networks [article]

Ismail Elezi
2020 arXiv   pre-print
Nevertheless, there has not been much research in exploiting contextual information in deep neural networks.  ...  For most part, the entire usage of contextual information has been limited to recurrent neural networks.  ...  identity anonymization generative adversarial networks [114] ; IEEE/CVF Com-The following paper was done after the thesis' submission, with the author having a secondary role.  ... 
arXiv:2006.11706v2 fatcat:bzaghxubbzcftdsvv4ogk3tfxe