StyleGAN as a Utility-Preserving Face De-identification Method

  title={StyleGAN as a Utility-Preserving Face De-identification Method},
  author={Seyyed Mohammad Sadegh Moosavi Khorzooghi and Shirin Nilizadeh},
Several face de-identification methods have been proposed to preserve users’ privacy by obscuring their faces. These methods, however, can degrade the quality of photos, and they usually do not preserve the utility of faces, e.g., their age, gender, pose, and facial expression. Recently, advanced generative adversarial network models, such as StyleGAN [20], have been proposed, which generate realistic, high-quality imaginary faces. In this paper, we investigate the use of StyleGAN in generating… 

Figures and Tables from this paper

Towards a Controllable and Reversible Privacy Protection System for Facial Images through Enhanced Multi-Factor Modifier Networks

This work proposes a new controllable and reversible privacy protection system that can automatically and stably anonymize and de-anonymize face images with one neural network and provide strong security protection with multi-factor identification solutions.



AnonymousNet: Natural Face De-Identification With Measurable Privacy

  • Tao LiLei Lin
  • Computer Science
    2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW)
  • 2019
Not only does this paper achieve the state-of-the-arts in terms of image quality and attribute prediction accuracy, it is also the first to show that facial privacy is measurable, can be factorized, and accordingly be manipulated in a photo-realistic fashion to fulfill different requirements and application scenarios.

A Utility-Preserving GAN for Face Obscuration

A utility-preserving generative model that is able to provide an effective face obscuration, while preserving facial utility, is proposed, and it is shown that the proposed method achieves the best performance in terms of obscuration and utility preservation.

Face Deidentification with Generative Deep Neural Networks

This work presents a novel face deidentification pipeline, which ensures anonymity by synthesizing artificial surrogate faces using generative neural networks (GNNs) to deidentify subjects in images or video, while preserving non-identity-related aspects of the data and consequently enabling data utilization.

Integrating Utility into Face De-identification

A new algorithm, k-Same-Select, is introduced, which is a formal privacy protection schema based on k-anonymity that provably protects privacy and preserves data utility and empirically validate the findings through evaluations on the FERET database.

Model-Based Face De-Identification

It is shown in extensive experiments that pixelation and blurring offers very poor privacy protection while significantly distorting the data and a novel framework for de-identifying facial images is introduced, which combines a model-based face image parameterization with a formal privacy protection model.

CIAGAN: Conditional Identity Anonymization Generative Adversarial Networks

CIAGAN, a model for image and video anonymization based on conditional generative adversarial networks, is proposed and developed, able to remove the identifying characteristics of faces and bodies while producing high-quality images and videos that can be used for any computer vision task, such as detection or tracking.

DeepPrivacy: A Generative Adversarial Network for Face Anonymization

A novel architecture which is able to automatically anonymize faces in images while retaining the original data distribution is proposed, based on a conditional generative adversarial network, which generates highly realistic faces with a seamless transition between the generated face and the existing background.

A Hybrid Model for Identity Obfuscation by Face Replacement

This work proposes a new hybrid approach to obfuscate identities in photos by head replacement that improves over the previous state of the art in obfuscation rate while preserving a higher similarity to the original image content.

Learning to Anonymize Faces for Privacy Preserving Action Detection

A new principled approach for learning a video face anonymizer that performs a pixel-level modification to anonymize each person's face, with minimal effect on action detection performance.

Natural and Effective Obfuscation by Head Inpainting

This work proposes a novel head inpainting obfuscation technique that generates realistic person images, while achieving superior obfuscation performance against automatic person recognizers.