Abstract: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, i.e., their age, gender, pose, and facial expression. Recently, GANs, such as StyleGAN, have been proposed, which generate realistic, high-quality imaginary faces. In this paper, we investigate the use of StyleGAN in generating de-identified faces through style mixing. We examined this de-identification method for preserving utility and privacy by implementing several face detection, verification, and identification attacks and conducting a user study. The results from our extensive experiments, human evaluation, and comparison with two state-of-the-art methods, i.e., CIAGAN and DeepPrivacy, show that StyleGAN performs on par or better than these methods, preserving users' privacy and images' utility. In particular, the results of the machine learning-based experiments show that StyleGAN0-4 preserves utility better than CIAGAN and DeepPrivacy while preserving privacy at the same level. StyleGAN0-3 preserves utility at the same level while providing more privacy. In this paper, for the first time, we also performed a carefully designed user study to examine both privacy and utility-preserving properties of StyleGAN0-3, 0-4, and 0-5, as well as CIAGAN and DeepPrivacy from the human observers' perspectives. Our statistical tests showed that participants tend to verify and identify StyleGAN0-5 images more easily than DeepPrivacy images. All the methods but StyleGAN0-5 had significantly lower identification rates than CIAGAN. Regarding utility, as expected, StyleGAN0-5 performed significantly better in preserving some attributes. Among all methods, on average, participants believe gender has been preserved the most while naturalness has been preserved the least.

What problem does this paper attempt to address?

The paper primarily explores how to utilize StyleGAN as a practical method for facial de-identification, aiming to protect personal privacy while maintaining the utility of the images. Specifically, the paper attempts to address the following issues: 1. **Balancing Privacy Protection and Utility**: Existing facial de-identification methods can protect privacy but often reduce image quality or lose key facial attributes (such as age, gender, pose, and expression). The paper proposes using StyleGAN to generate de-identified facial images to retain these useful pieces of information while protecting privacy. 2. **Exploration of StyleGAN Application**: Although StyleGAN is designed to generate high-quality synthetic facial images, this paper explores whether it can serve as an effective facial de-identification tool. The study generates new de-identified facial images by mixing the style features of the target face and auxiliary faces. 3. **Evaluation and Comparison**: The paper conducts a series of experiments to evaluate the effectiveness and robustness of StyleGAN as a facial de-identification method and compares its results with two state-of-the-art facial de-identification methods (CIAGAN and DeepPrivacy). The evaluation includes the ability to protect privacy under automated machine learning attacks and the retention of utility as assessed by the Face++ system. 4. **User Study**: To further validate the privacy protection and utility of StyleGAN from the perspective of human observers, the paper also conducts a carefully designed user study, marking the first time this approach has been used in such research. In summary, the paper aims to address the key challenge in facial de-identification technology: how to effectively protect personal privacy while maintaining the utility of the images as much as possible, especially in AI and computer vision applications. By using StyleGAN and its style-mixing capabilities, the researchers seek to find a method that can both protect privacy and retain key facial attributes, and they comprehensively evaluate its performance.

StyleGAN as a Utility-Preserving Face De-identification Method

Privacy-Protective-GAN for Privacy Preserving Face De-Identification

StyleID: Identity Disentanglement for Anonymizing Faces

A Hybrid Model for Natural Face De-Identiation with Adjustable Privacy

How to Boost Face Recognition with StyleGAN?

StyleIPSB: Identity-Preserving Semantic Basis of StyleGAN for High Fidelity Face Swapping

PrivacyGAN: robust generative image privacy

Privacy preservation through facial de-identification with simultaneous emotion preservation

Face De-Identification Using Face Caricature

Effective De-identification Generative Adversarial Network for Face Anonymization

StyleFace: Towards Identity-Disentangled Face Generation on Megapixels.

FICGAN: Facial Identity Controllable GAN for De-identification

Privacy preservation for image data: A GAN‐based method

High similarity controllable face anonymization based on dynamic identity perception

Subverting Privacy-Preserving GANs: Hiding Secrets in Sanitized Images

Face De-identification: State-of-the-art Methods and Comparative Studies

G2Face: High-Fidelity Reversible Face Anonymization via Generative and Geometric Priors

Semantic-aware Deidentification Generative Adversarial Networks for Identity Anonymization

Few-shots Portrait Generation with Style Enhancement and Identity Preservation

Gender-Adversarial Networks for Face Privacy Preserving