Attribute-preserving Face Dataset Anonymization via Latent Code Optimization

Abstract

This work addresses the problem of anonymizing the identity of faces in a dataset of images, such that the privacy of those depicted is not violated, while at the same time the dataset is useful for downstream task such as for training machine learning models. To the best of our knowledge, we are the first to explicitly address this issue and deal with two major drawbacks of the existing state-of-the-art approaches, namely that they (i) require the costly training of additional, purpose-trained neural networks, and/or (ii) fail to retain the facial attributes of the original images in the anonymized counterparts, the preservation of which is of paramount importance for their use in downstream tasks. We accordingly present a task-agnostic anonymization procedure that directly optimizes the images' latent representation in the latent space of a pre-trained GAN. By optimizing the latent codes directly, we ensure both that the identity is of a desired distance away from the original (with an identity obfuscation loss), whilst preserving the facial attributes (using a novel feature-matching loss in FaRL's [49] deep feature space). We demonstrate through a series of both qualitative and quantitative experiments that our method is capable of anonymizing the identity of the images whilstcrucially-better-preserving the facial attributes. We make the code and the pre-trained models publicly available at: https://github.com/chi0tzp/FALCO.

PICLAnony: Controlled Anonymization of Face Attributes Based on Parametric Imitation Comparison Learning

“…This means that we can generate anonymized faces that satisfy different privacy requirements, avoiding the anonymized images that lead to the exposure of sensitive behavioral intentions due to the fact that the expressions, poses, and lighting are the same as those of the source images. (Liu et al (2015)) in comparison to CIAGAN (Maximov et al (2020)), AnonyGAN(Dall'Asen et al ( 2022)) and FALCO (Barattin et al (2023)). Where a, b, c, d, and e represent groups of anonymized images of the same image under different methods.…”

“…5. FALCO (Barattin et al (2023)). FALCO is a recent GAN based method for face anonymization while preserving attributes, which quantitatively evaluates the degree of attribute preservation of anonymized faces.…”

“…Although the method effectively blurs the identity-related information of a face, it does not focus on the preservation of facial attributes. Based on this, FALCO (Barattin et al (2023)) evaluated the preservation of facial attributes while achieving anonymity, quantifying the degree of preservation. However, most of these anonymization solutions protect identity information based on the original image.…”

“…Fig.7Anonymization results on CelebA HQ(Liu et al (2015)) in comparison to CIAGAN(Maximov et al (2020)), AnonyGAN(Dall'Asen et al (2022)) and FALCO(Barattin et al (2023)). Where a, b, c, d, and e represent groups of anonymized images of the same image under different methods.…”