2022-06-26 更新
JPEG Compression-Resistant Low-Mid Adversarial Perturbation against Unauthorized Face Recognition System
Authors:Jiaming Zhang, Qi Yi, Jitao Sang
It has been observed that the unauthorized use of face recognition system raises privacy problems. Using adversarial perturbations provides one possible solution to address this issue. A critical issue to exploit adversarial perturbation against unauthorized face recognition system is that: The images uploaded to the web need to be processed by JPEG compression, which weakens the effectiveness of adversarial perturbation. Existing JPEG compression-resistant methods fails to achieve a balance among compression resistance, transferability, and attack effectiveness. To this end, we propose a more natural solution called low frequency adversarial perturbation (LFAP). Instead of restricting the adversarial perturbations, we turn to regularize the source model to employing more low-frequency features by adversarial training. Moreover, to better influence model in different frequency components, we proposed the refined low-mid frequency adversarial perturbation (LMFAP) considering the mid frequency components as the productive complement. We designed a variety of settings in this study to simulate the real-world application scenario, including cross backbones, supervisory heads, training datasets and testing datasets. Quantitative and qualitative experimental results validate the effectivenss of proposed solutions.
PDF
论文截图
The Gender Gap in Face Recognition Accuracy Is a Hairy Problem
Authors:Aman Bhatta, Vítor Albiero, Kevin W. Bowyer, Michael C. King
It is broadly accepted that there is a “gender gap” in face recognition accuracy, with females having higher false match and false non-match rates. However, relatively little is known about the cause(s) of this gender gap. Even the recent NIST report on demographic effects lists “analyze cause and effect” under “what we did not do”. We first demonstrate that female and male hairstyles have important differences that impact face recognition accuracy. In particular, compared to females, male facial hair contributes to creating a greater average difference in appearance between different male faces. We then demonstrate that when the data used to estimate recognition accuracy is balanced across gender for how hairstyles occlude the face, the initially observed gender gap in accuracy largely disappears. We show this result for two different matchers, and analyzing images of Caucasians and of African-Americans. These results suggest that future research on demographic variation in accuracy should include a check for balanced quality of the test data as part of the problem formulation. To promote reproducible research, matchers, attribute classifiers, and datasets used in this research are/will be publicly available.
PDF
论文截图
3D Face Morphing Attacks: Generation, Vulnerability and Detection
Authors:Jag Mohan Singh, Raghavendra Ramachandra
Face Recognition systems (FRS) have been found vulnerable to morphing attacks, where the morphed face image is generated by blending the face images from contributory data subjects. This work presents a novel direction towards generating face morphing attacks in 3D. To this extent, we have introduced a novel approach based on blending the 3D face point clouds corresponding to the contributory data subjects. The proposed method will generate the 3D face morphing by projecting the input 3D face point clouds to depth-maps \& 2D color images followed by the image blending and wrapping operations performed independently on the color images and depth maps. We then back-project the 2D morphing color-map and the depth-map to the point cloud using the canonical (fixed) view. Given that the generated 3D face morphing models will result in the holes due to a single canonical view, we have proposed a new algorithm for hole filling that will result in a high-quality 3D face morphing model. Extensive experiments are carried out on the newly generated 3D face dataset comprised of 675 3D scans corresponding to 41 unique data subjects. Experiments are performed to benchmark the vulnerability of automatic 2D and 3D FRS and human observer analysis. We also present the quantitative assessment of the quality of the generated 3D face morphing models using eight different quality metrics. Finally, we have proposed three different 3D face Morphing Attack Detection (3D-MAD) algorithms to benchmark the performance of the 3D MAD algorithms.
PDF The paper is currently under review at IEEE Transactions on Biometrics, Behavior and Identity Science
论文截图