Accurate and Robust Neural Networks for Security Related Applications Exampled by Face Morphing Attacks

Seibold, Clemens; Samek, Wojciech; Hilsmann, Anna; Eisert, Peter

doi:10.48550/arxiv.1806.04265

Cited by 5 publications

(9 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The landmark based image morphing technique [23,9,38,12] considers the geometry of two bona fide subjects for generating a morphed sample. Every pixel location in both genuine images is warped to preserve the correspondence in the resulting morphed sample, and a convex combination of the warped pixels cross-dissolve the warped pixels in the real images to synthesize that pixel location in the final morphed sample.…”

Section: Morph Generationmentioning

confidence: 99%

“…Some research considers hand-crafted descriptors to detect morphed samples [11,29,34,28], where the fusion of different features has proven to be compelling for morph detection [34,32]. Deep morph detectors utilize the power of convolutional neural networks to detect morphed samples accurately [14,30,37,38,36,22]. Several studies investigate denoising methods for the task of morph attack detection.…”

Section: Morph Detectionmentioning

confidence: 99%

See 1 more Smart Citation

Morph Detection Enhanced by Structured Group Sparsity

Aghdaie¹,

Chaudhary²,

Soleymani³

et al. 2021

Preprint

View full text Add to dashboard Cite

In this paper, we consider the challenge of face morphing attacks, which substantially undermine the integrity of face recognition systems such as those adopted for use in border protection agencies. Morph detection can be formulated as extracting fine-grained representations, where local discriminative features are harnessed for learning a hypothesis. To acquire discriminative features at different granularity as well as a decoupled spectral information, we leverage wavelet domain analysis to gain insight into the spatialfrequency content of a morphed face. As such, instead of using images in the RGB domain, we decompose every image into its wavelet sub-bands using 2D wavelet decomposition and a deep supervised feature selection scheme is employed to find the most discriminative wavelet sub-bands of input images. To this end, we train a Deep Neural Network (DNN) morph detector using the decomposed wavelet subbands of the morphed and bona fide images. In the training phase, our structured group sparsity-constrained DNN picks the most discriminative wavelet sub-bands out of all the sub-bands, with which we retrain our DNN, resulting in a precise detection of morphed images when inference is achieved on a probe image. The efficacy of our deep morph detector which is enhanced by structured group lasso is validated through experiments on three facial morph image databases, i.e., VISAPP17, LMA, and MorGAN.

show abstract

Section: Morph Generationmentioning

confidence: 99%

Section: Morph Detectionmentioning

confidence: 99%

Morph Detection Enhanced by Structured Group Sparsity

Aghdaie¹,

Chaudhary²,

Soleymani³

et al. 2021

Preprint

View full text Add to dashboard Cite

show abstract

“…Facial morph generation techniques are categorized into two types, i.e., landmark-based morphing [5,8,19,31], and GAN-based morphing [5,37]. In the landmark-based morphing attack, appearance of a resulting morphed image is associated with that of two underlying subject's bona fide face images, while geometric locations of its landmarks are the average of the corresponding landmarks in the two bona fide images [32].…”

Section: Morph Generationmentioning

confidence: 99%

Attention Aware Wavelet-based Detection of Morphed Face Images

Aghdaie¹,

Chaudhary²,

Soleymani³

et al. 2021

Preprint

View full text Add to dashboard Cite

Morphed images have exploited loopholes in the face recognition checkpoints, e.g., Credential Authentication Technology (CAT), used by Transportation Security Administration (TSA), which is a non-trivial security concern. To overcome the risks incurred due to morphed presentations, we propose a wavelet-based morph detection methodology which adopts an end-to-end trainable soft attention mechanism . Our attention-based deep neural network (DNN) focuses on the salient Regions of Interest (ROI) which have the most spatial support for morph detector decision function, i.e, morph class binary softmax output. A retrospective of morph synthesizing procedure aids us to speculate the ROI as regions around facial landmarks , particularly for the case of landmark-based morphing techniques. Moreover, our attention-based DNN is adapted to the wavelet space, where inputs of the network are coarse-to-fine spectral representations, 48 stacked wavelet sub-bands to be exact. We evaluate performance of the proposed framework using three datasets, VISAPP17, LMA, and MorGAN. In addition, as attention maps can be a robust indicator whether a probe image under investigation is genuine or counterfeit, we analyze the estimated attention maps for both a bona fide image and its corresponding morphed image. Finally, we present an ablation study on the efficacy of utilizing attention mechanism for the sake of morph detection.

show abstract

“…One of the first efforts to study the generation of a morph image from two source images [13] has concluded that geometric alterations and digital beautification can cause an increase in the possibility of fooling recognition systems. Morph generation techniques can roughly be categorized into landmark-based [29,43,44] and generative models [10,45]. Landmark-based frameworks focus on detecting the landmarks in both the images, translating these points toward each other, and blending the two face images.…”

Section: Facial Morphingmentioning

confidence: 99%

“…Texture descriptors are the main feature extraction models for single image morph detection [51,47,38,37,34]. Recently, deep learning models have also been considered for this purpose [44,43,33]. The models mentioned can also be employed for differential morph detection when the extracted feature from the two images are compared [35,39,9].…”

Section: Morph Detectionmentioning

confidence: 99%

Mutual Information Maximization on Disentangled Representations for Differential Morph Detection

Soleymani¹,

Dabouei²,

Taherkhani³

et al. 2020

Preprint

View full text Add to dashboard Cite

In this paper, we present a novel differential morph detection framework, utilizing landmark and appearance disentanglement. In our framework, the face image is represented in the embedding domain using two disentangled but complementary representations. The network is trained by triplets of face images, in which the intermediate image inherits the landmarks from one image and the appearance from the other image. This initially trained network is further trained for each dataset using contrastive representations. We demonstrate that, by employing appearance and landmark disentanglement, the proposed framework can provide state-of-the-art differential morph detection performance. This functionality is achieved by the using distances in landmark, appearance, and ID domains. The performance of the proposed framework is evaluated using three morph datasets generated with different methodologies.

show abstract

Accurate and Robust Neural Networks for Security Related Applications Exampled by Face Morphing Attacks

Cited by 5 publications

References 0 publications

Morph Detection Enhanced by Structured Group Sparsity

Morph Detection Enhanced by Structured Group Sparsity

Attention Aware Wavelet-based Detection of Morphed Face Images

Mutual Information Maximization on Disentangled Representations for Differential Morph Detection

Contact Info

Product

Resources

About