Privacy-Preserving Face Recognition Using Random Frequency Components

TL;DR

This paper introduces PartialFace, a privacy-preserving face recognition method that conceals visual information by pruning low-frequency components and uses random frequency components during training and inference to balance privacy and accuracy.

Contribution

It proposes a novel approach to face recognition that enhances privacy by selectively removing visual information and employing random frequency components, addressing the privacy-accuracy dilemma.

Findings

PartialFace effectively balances privacy and recognition accuracy.

Training and inference on random frequency components improve privacy protection.

Extensive experiments validate the method's effectiveness.

Abstract

The ubiquitous use of face recognition has sparked increasing privacy concerns, as unauthorized access to sensitive face images could compromise the information of individuals. This paper presents an in-depth study of the privacy protection of face images' visual information and against recovery. Drawing on the perceptual disparity between humans and models, we propose to conceal visual information by pruning human-perceivable low-frequency components. For impeding recovery, we first elucidate the seeming paradox between reducing model-exploitable information and retaining high recognition accuracy. Based on recent theoretical insights and our observation on model attention, we propose a solution to the dilemma, by advocating for the training and inference of recognition models on randomly selected frequency components. We distill our findings into a novel privacy-preserving face recognition method, PartialFace. Extensive experiments demonstrate that PartialFace effectively balances privacy protection goals and recognition accuracy. Code is available at: https://github.com/Tencent/TFace.