Detecting and Mitigating Adversarial Perturbations for Robust Face Recognition

Goswami, Gaurav; Agarwal, Akshay; Ratha, Nalini K.; Singh, Richa; Vatsa, Mayank

doi:10.1007/s11263-019-01160-w

Cited by 99 publications

(47 citation statements)

References 55 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Goswami et al proposed a framework for face recognition based on deep neural networks and used a characteristic anomaly filter to detect singularities. Finally, after verification with public data, it was found that the model built by it had strong face recognition robustness [11]. Isogawa et al indicated that there was no denoising method capable of parameter adjustment in the DCNN model.…”

Section: Related Workmentioning

confidence: 99%

[Retracted] Analysis of Feature Extraction and Anti‐Interference of Face Image under Deep Reconstruction Network Algorithm

et al. 2021

View full text Add to dashboard Cite

In face recognition systems, highly robust facial feature representation and good classification algorithm performance can affect the effect of face recognition under unrestricted conditions. To explore the anti-interference performance of convolutional neural network (CNN) reconstructed by deep learning (DL) framework in face image feature extraction (FE) and recognition, in the paper, first, the inception structure in the GoogleNet network and the residual error in the ResNet network structure are combined to construct a new deep reconstruction network algorithm, with the random gradient descent (SGD) and triplet loss functions as the model optimizer and classifier, respectively, and it is applied to the face recognition in Labeled Faces in the Wild (LFW) face database. Then, the portrait pyramid segmentation and local feature point segmentation are applied to extract the features of face images, and the matching of face feature points is achieved using Euclidean distance and joint Bayesian method. Finally, Matlab software is used to simulate the algorithm proposed in this paper and compare it with other algorithms. The results show that the proposed algorithm has the best face recognition effect when the learning rate is 0.0004, the attenuation coefficient is 0.0001, the training method is SGD, and dropout is 0.1 (accuracy: 99.03%, loss: 0.0047, training time: 352 s, and overfitting rate: 1.006), and the algorithm proposed in this paper has the largest mean average precision compared to other CNN algorithms. The correct rate of face feature matching of the algorithm proposed in this paper is 84.72%, which is higher than LetNet-5, VGG-16, and VGG-19 algorithms, the correct rates of which are 6.94%, 2.5%, and 1.11%, respectively, but lower than GoogleNet, AlexNet, and ResNet algorithms. At the same time, the algorithm proposed in this paper has a faster matching time (206.44 s) and a higher correct matching rate (88.75%) than the joint Bayesian method, indicating that the deep reconstruction network algorithm proposed in this paper can be used in face image recognition, FE, and matching, and it has strong anti-interference.

show abstract

Section: Related Workmentioning

confidence: 99%

[Retracted] Analysis of Feature Extraction and Anti‐Interference of Face Image under Deep Reconstruction Network Algorithm

et al. 2021

View full text Add to dashboard Cite

show abstract

“…Adding two external models to the classifier network, they explored two feature squeezing approaches by (1) decreasing the color bit depth of each pixel and (2) spatial smoothing. Goswami et al [14] expressed that this approach is simple and operative for highresolution images with detailed data; however, it may not be operational for low resolution cropped faces frequently used in FR settings. In [114], an open-source Python-based toolbox, termed as SmartBox, is proposed to benchmark the function of adversarial attack detection and mitigation algorithms against FR models.…”

Section: ) Supplementing External Networkmentioning

confidence: 99%

Adversarial Attacks Against Face Recognition: A Comprehensive Study

2021

View full text Add to dashboard Cite

Face recognition (FR) systems have demonstrated reliable verification performance, suggesting suitability for real-world applications ranging from photo tagging in social media to automated border control (ABC). In an advanced FR system with deep learning-based architecture, however, promoting the recognition efficiency alone is not sufficient, and the system should also withstand potential kinds of attacks. Recent studies show that (deep) FR systems exhibit an intriguing vulnerability to imperceptible or perceptible but natural-looking adversarial input images that drive the model to incorrect output predictions. In this article, we present a comprehensive survey on adversarial attacks against FR systems and elaborate on the competence of new countermeasures against them. Further, we propose a taxonomy of existing attack and defense methods based on different criteria. We compare attack methods on the orientation, evaluation process, and attributes, and defense approaches on the category. Finally, we discuss the challenges and potential research direction.

show abstract

“…On the other hand, in recent years, deep neural networks (DNN) attracted many researchers. However, there are some drawbacks as reported in Goswami et al ( 2019 ) that the DNN architecture-based methods are mostly a black box method since it is not easy to mathematically formulate the functions that are learned within its many layers of representation. They have examined the robustness of DNNs for face recognition and reported that the performance of deep learning-based face recognition algorithms significantly suffers in the presence of vulnerabilities and adversaries.…”

Section: Related Workmentioning

confidence: 99%

A hybrid method for biometric authentication-oriented face detection using autoregressive model with Bayes Backpropagation Neural Network

Vasanthi

Seetharaman

2021

Soft Comput

View full text Add to dashboard Cite

This paper proposes a novel method, which is coined as ARBBPNN, for biometric-oriented face detection, based on autoregressive model with Bayes backpropagation neural network (BBPNN). Firstly, the given input colour key face image is modelled to HSV and YCbCr models. A hybrid model, called HS-YCbCr, is formulated based on the HSV and YCbCr models. The submodel, H, is divided into various sliding windows of size, 3 9 3. The model parameters are estimated for the window using the BBPNN. Based on the model coefficients, autocorrelation coefficients (ACCs) are computed. An autocorrelation test tests the significance of the ACCs. If the ACC passes the test, then it is inferred that the small image region, viz. the window, represents the texture and it is treated as the texture feature. Otherwise, it is regarded as structure, which is treated as the shape feature. The texture and shape features are formulated as feature vectors (FV) separately, and they are combined into a single FV. This process is performed for all colour submodels. The FVs of the submodels are combined into a single holistic vector, which is treated as the FV of the key face image. The key FV has twenty feature elements. The similarity of the key and target face images is examined, based on the key and target FVs, by deploying multivariate parametric statistical tests. If the FVs of the key and target images pass the tests, then it is concluded that the key and target face images are the same; otherwise, they are regarded as different. The GT, FSW, Pointing'04, and BioID datasets are considered for the experiments. In addition to the above datasets, we have constructed a dataset with face images collected from Google, and many images captured through a digital camera. It is also subjected to the experiment. The obtained recognition results show that the proposed ARBBPNN method outperforms the existing methods.

show abstract

Detecting and Mitigating Adversarial Perturbations for Robust Face Recognition

Cited by 99 publications

References 55 publications

[Retracted] Analysis of Feature Extraction and Anti‐Interference of Face Image under Deep Reconstruction Network Algorithm

[Retracted] Analysis of Feature Extraction and Anti‐Interference of Face Image under Deep Reconstruction Network Algorithm

Adversarial Attacks Against Face Recognition: A Comprehensive Study

A hybrid method for biometric authentication-oriented face detection using autoregressive model with Bayes Backpropagation Neural Network

Contact Info

Product

Resources

About