Face Recognition Classification Research Articles

Beyond the visible: A survey on cross-spectral face recognition

Cross-spectral face recognition (CFR) refers to recognizing individuals using face images stemming from different spectral bands, such as infrared versus visible. While CFR is inherently more challenging than classical face recognition due to significant variation in facial appearance caused by the modality gap, it is useful in many scenarios including night-vision biometrics and detecting presentation attacks. Recent advances in deep neural networks (DNNs) have resulted in significant improvement in the performance of CFR systems. Given these developments, the contributions of this survey are three-fold. First, we provide an overview of CFR, by formalizing the CFR problem and presenting related applications. Secondly, we discuss the appropriate spectral bands for face recognition and discuss recent CFR methods, placing emphasis on deep neural networks. In particular we describe techniques that have been proposed to extract and compare heterogeneous features emerging from different spectral bands. We also discuss the datasets that have been used for evaluating CFR methods. Finally, we discuss the challenges and future lines of research on this topic.

Performance Evaluation of Some Selected Classification Algorithms in a Facial Recognition System

Facial Recognition (FR) has been an active area of research and has diverse applicable environment, it continues to be a challenging research topic. With the development of image processing and pattern recognition technology, there are many challenges in machine learning to select the appropriate classification algorithms, most especially in the area of classification of extracted features to have low classification time, high sensitivity and accuracy of the classification algorithms, so it is very important to explore the performance of different algorithms in image classification. The three selected supervised learning classification algorithms: Learning Vector Quantization (LVQ), Relevance Vector Machine (RVM), and Support Vector Machine (SVM) performance were evaluated so as to know the most effective out of the selected algorithms for facial images classification. The development of the system has four stages, the first stage is image acquisition and 180 images were taken by digital camera under same illumination and light colour background. The second stage is pre-processing to improve the images data by suppressing unwilling distortion; grayscale and normalization were used for image pre-processing. The third stage is feature extraction; Discrete Cosine Transform (DCT) is adopted for this purpose. While the fourth stage is face recognition classification, Receiver Operating Characteristics (ROC) was used to test the performance of each the three algorithms. However the Learning Vector Quantization algorithm, Relevance Vector Machine and Support Vector Machine performance have not been compared together to the most effective out of the three algorithms in term of False Positive Rate, Sensitivity, Specificity, Precision, Accuracy and Computation Time. Hence, this work evaluated the performance of the Learning Vector Quantization; Relevance Vector Machine and Support Vector Machine classification algorithms in facial recognition system and Support Vector Machine outwit the other two algorithms in facial recognition in term of specificity, recognition time and recognition accuracy at different threshold.

De-occlusion and recognition of frontal face images: a comparative study of multiple imputation methods

AbstractIncreasingly, automatic face recognition algorithms have become necessary with the development and extensive use of face recognition technology, particularly in the era of machine learning and artificial intelligence. However, the presence of unconstrained environmental conditions degrades the quality of acquired face images and may deteriorate the performance of many classical face recognition algorithms. Due to this backdrop, many researchers have given considerable attention to image restoration and enhancement mechanisms, but with minimal focus on occlusion-related and multiple-constrained problems. Although occlusion robust face recognition modules, via sparse representation have been explored, they require a large number of features to achieve correct computations and to maximize robustness to occlusions. Therefore, such an approach may become deficient in the presence of random occlusions of relatively moderate magnitude. This study assesses the robustness of Principal Component Analysis and Singular Value Decomposition using Discrete Wavelet Transformation for preprocessing and city block distance for classification (DWT-PCA/SVD-L1) face recognition module to image degradations due to random occlusions of varying magnitudes (10% and 20%) in test images acquired with varying expressions. Numerical evaluation of the performance of the DWT-PCA/SVD-L1 face recognition module showed that the use of the de-occluded faces for recognition enhanced significantly the performance of the study recognition module at each level (10% and 20%) of occlusion. The algorithm attained the highest recognition rate of 85.94% and 78.65% at 10% and 20% occlusions respectively, when the MICE de-occluded face images were used for recognition. With the exception of Entropy where MICE de-occluded face images attained the highest average value, the MICE and RegEM result in images of similar quality as measured by their Absolute mean brightness error (AMBE) and peak signal to noise ratio (PSNR). The study therefore recommends MICE as a suitable imputation mechanism for de-occlusion of face images acquired under varying expressions.

VGG16 based on dilated convolution for face recognition

Face recognition has wide applications in fields such as security systems, biometrics, and human-computer interaction. However, traditional face recognition methods face challenges in capturing details and reducing model complexity. To address these issues, this paper proposes a new method based on VGG16, which improves recognition accuracy and reduces parameter quantity by introducing dilated convolution and parameter pruning. First, the hole convolution is introduced to expand the Receptive field and capture more details to enhance the ability of the model in distinguishing facial features. Next, parameter pruning is applied to reduce redundant parameters, optimize model structure, and improve computational efficiency. This article conducted experimental evaluation on the classic face recognition dataset CK+ dataset. The results show that the proposed method is significantly superior to the traditional VGG16 model in terms of recognition accuracy. At the same time, the use of pruning technology significantly reduces the number of parameters in the model and improves computational efficiency. The experimental outcomes conclusively validate the effectiveness and feasibility of the proposed method.

Which common MRI research sequences need de‐facing before research data sharing?

AbstractBackgroundWe have previously shown that 3D T1‐weighted and 3D T2‐FLAIR brain MRI can potentially be re‐identified using face recognition, but this has not been tested for other MRI sequences commonly used in imaging studies of aging and dementia. Here we compare match rates of automated face recognition to re‐identify 3D T1‐weighted, 3D T2‐FLAIR, 3D T2‐weighted, 2D Axial T2‐FLAIR, multi‐echo GRE (T2*), diffusion MRI, functional MRI, ASL, and T2* GRE images.MethodWe used a dataset of 182 Mayo Clinic participants each with MRI and face photographs. MRIs used Siemens and GE vendor‐product sequences similar to ADNI3 and ADNI4 protocols. We created face reconstructions (“basic”, with minimal preprocessing, and “advanced” that automatically filled missing face regions using a template) from each MRI sequence, and we trained a Microsoft Azure PersonGroup face recognition classifier to recognize these MRI‐based faces. We then queried the classifier with five face photographs for each participant, and we recorded how often the classifier chose the correct MRI as the top‐ranked match. For image types with match rates > = 10%, we then repeated the experiment after de‐facing each MRI with mri_reface, our leading software that replaces faces in each image (“re‐faces”) with an average face.ResultResults are presented in Table 1. 3D T2‐FLAIR, 3D T1‐weighted, and 3D T2‐weighted images were matched at 96‐98%, and mri_reface reduced all to 8%. Axial T2‐FLAIR and ME‐GRE were matched at 32‐45%, and mri_reface reduced both to 1%. T2*‐GRE matched at 10%, and mri_reface reduced this to 1%. ASL matched at 8%, while dMRI and fMRI matched at only 0‐2%.ConclusionAmong current‐generation vendor‐product sequences, we recommend that de‐facing be applied before research data sharing of T1‐weighted, T2‐weighted, T2‐FLAIR, T2*, and ME‐GRE images. Minimal match rates for current‐generation product fMRI, ASL, and fMRI images suggest that they present low risk of re‐identification and can be shared without de‐facing, but this conclusion should be re‐evaluated if they are acquired without fat‐suppression, with a larger field of view, or if newer developments reduce the current levels of artifacts and distortion around the face.

Which common MRI research sequences need de‐facing before research data sharing?

AbstractBackgroundWe have previously shown that 3D T1‐weighted and 3D T2‐FLAIR brain MRI can potentially be re‐identified using face recognition, but this has not been tested for other MRI sequences commonly used in imaging studies of aging and dementia. Here we compare match rates of automated face recognition to re‐identify 3D T1‐weighted, 3D T2‐FLAIR, 3D T2‐weighted, 2D Axial T2‐FLAIR, multi‐echo GRE (T2*), diffusion MRI, functional MRI, ASL, and T2* GRE images.MethodWe used a dataset of 182 Mayo Clinic participants each with MRI and face photographs. MRIs used Siemens and GE vendor‐product sequences similar to ADNI3 and ADNI4 protocols. We created face reconstructions (“basic”, with minimal preprocessing, and “advanced” that automatically filled missing face regions using a template) from each MRI sequence, and we trained a Microsoft Azure PersonGroup face recognition classifier to recognize these MRI‐based faces. We then queried the classifier with five face photographs for each participant, and we recorded how often the classifier chose the correct MRI as the top‐ranked match. For image types with match rates > = 10%, we then repeated the experiment after de‐facing each MRI with mri_reface, our leading software that replaces faces in each image (“re‐faces”) with an average face.ResultResults are presented in Table 1. 3D T2‐FLAIR, 3D T1‐weighted, and 3D T2‐weighted images were matched at 96‐98%, and mri_reface reduced all to 8%. Axial T2‐FLAIR and ME‐GRE were matched at 32‐45%, and mri_reface reduced both to 1%. T2*‐GRE matched at 10%, and mri_reface reduced this to 1%. ASL matched at 8%, while dMRI and fMRI matched at only 0‐2%.ConclusionAmong current‐generation vendor‐product sequences, we recommend that de‐facing be applied before research data sharing of T1‐weighted, T2‐weighted, T2‐FLAIR, T2*, and ME‐GRE images. Minimal match rates for current‐generation product fMRI, ASL, and fMRI images suggest that they present low risk of re‐identification and can be shared without de‐facing, but this conclusion should be re‐evaluated if they are acquired without fat‐suppression, with a larger field of view, or if newer developments reduce the current levels of artifacts and distortion around the face.

Deep learning framework for real-time face recognition using multimodal facial expression with feature optimization and hybrid classification

Smile detection plays key role in recognizing human facial expressions, especially in inter-personal relations. In modern biometrics and security applications, face recognition is widely used. In real-life applications, automated face recognition has always been challenging because people express themselves differently. Real-world scenarios present many challenges, such as low-quality images, temporal variations, and disguises that alter facial characteristics. Face recognition has been accomplished using deep learning frameworks, but there are still several challenges to overcome, such as expression variations and lack of training data. In these methods, margins are used to enforce intra-class compactness and inter-class discrepancy to prevent overfitting. Multimodal facial expression detection is used for real-time face recognition to solve the aforementioned problems. First, we utilizes the Prewitt edge detector for eyes, nose, mouth, and brow detection, and design a deep transfer learning model to extract deep features from face images. Next, feature optimization is done by harpy eagle search optimization (HESO) to selects optimal best among multiple features which reduce data dimensionality problem. Then, we employ a semi-region based convolutional neural network (SR-CNN) to analyzes the facial expressions from multimodalities such as eyes (blinking), nose (shapes), mouth (smiling), and brow (shapes). The analyzed facial expressions and corresponding input images are fed into the benchmark SVM classifier for face recognition. The facial expression based face recognition framework is most suited for real-time scenario like video surveillance. Finally, we validate the performance our framework using benchmark multi-modal Cohn-Kanade and real-time 5671 person images datasets. The simulation results show that the proposed framework can effectively suppress data dimensionality issues and achieve high recognition rate compare to state-of-art frameworks.

Gravitational Deep Convoluted Stacked Kernel Extreme Learning Based Classification for Face Recognition

In recent times, researchers have designed several deep learning (DL) algorithms and specifically face recognition (FR) made an extensive crossover. Deep Face Recognition systems took advantage of the hierarchical framework of the DL algorithms to learn discriminative face characterization. However, when handling severe occlusions in a face, the execution of present-day methods reduces appreciably. Several prevailing works regard that, when face recognition is taken into consideration, affinity materializes to be a pivotal recognition feature. However, the rate of affinity changes when the face image for recognition is found to be illuminated, and occluded, with changes in the age of the subject. Motivated by these issues, in this work a novel method called Gravitational Deep Convoluted Stacked Kernel Extreme Learning-based (GDC-SKEL) classification for face recognition is proposed for human face recognition problems in frontal views with varying age, illumination, and occlusion. First, with the face images provided as input, Gravitational Center Loss-based Face Alignment model is proposed to minimize the intra-class difference, which can overcome the influence of occlusion in face images. Second, Deep Convoluted Tikhonov Regularization-based Facial Region Feature extraction is applied to the occlusion-removed face images. Here, by employing the Convoluted Tikhonov Regularization function, salient features are said to be extracted with an age-invariant representation. Finally, Stacked Kernel Extreme Learning-based Classification is designed. The extracted features are given to the Stacked Kernel Extreme Learning-based Classification and to identify testing samples Stacked Kernel is utilized. The performance of GDC-SKEL is evaluated on Cross-Age Celebrity Dataset. Experimental results are compared with other state-of-the-art classifiers in terms of face recognition accuracy, face recognition time, PSNR, and False Positive Rate which shows the effectiveness of the proposed GDC-SKEL classifier.

Retracted: Research on Face Recognition Classification Based on Improved GoogleNet

Retracted: Research on Face Recognition Classification Based on Improved GoogleNet

Oppositional Grass Hopper Optimization with Fuzzy Classifier for Face Recognition from Video Database

Oppositional Grass Hopper Optimization with Fuzzy Classifier for Face Recognition from Video Database

SAFE: Security Door Lock System Using Haar-Cascade and LBPH Method

Door lock provides numerous benefits and has become indispensable in daily life as it acts as a security guard to prevent burglars and protect home belonging safely. The unlock methods of existing door lock system widely use keys and thumbprint involve touching the object may cause the spread of COVID-19. In this paper, a computer vision based security door lock system using Raspberry Pi (called SAFE) is proposed. Haar-Cascade classifier is employed as face detection classifier, while Local Binary Pattern Histogram (LBPH) is proposed as face recognition classifier. Recognition result is processed based on the usage of user to provide insights of SAFE. The accuracy of SAFE using pre-trained LBPH classifier achieves average of 86% based on the data obtained. The recognition speed outperforms existing work using principal component analysis and eigenfaces.

Group Class Residual ℓ1-Minimization on Random Projection Sparse Representation Classifier for Face Recognition

Sparse Representation-based Classification (SRC) has been seen to be a reliable Face Recognition technique. The ℓ1 Bayesian based on the Lasso algorithm has proven to be most effective in class identification and computation complexity. In this paper, we revisit classification algorithm and then recommend the group-based classification. The proposed modified algorithm, which is called as Group Class Residual Sparse Representation-based Classification (GCR-SRC), extends the coherency of the test sample to the whole training samples of the identified class rather than only to the nearest one of the training samples. Our method is based on the nearest coherency between a test sample and the identified training samples. To reduce the dimension of the training samples, we choose random projection for feature extraction. This method is selected to reduce the computational cost without increasing the algorithm’s complexity. From the simulation result, the reduction factor (ρ) 64 can achieve a maximum recognition rate about 10% higher than the SRC original using the downscaling method. Our proposed method’s feasibility and effectiveness are tested on four popular face databases, namely AT&T, Yale B, Georgia Tech, and AR Dataset. GCR-SRC and GCR-RP-SRC achieved up to 4% more accurate than SRC random projection with class-specific residuals. The experiment results show that the face recognition technology based on random projection and group-class-based not only reduces the dimension of the face data but also increases the recognition accuracy, indicating that it is a feasible method for face recognition.

Comparative analysis of classifiers for face recognition on image fragments identified by the FaceNet neural network

As a result of the analysis of the literature, the based methods of face recognition on fragments of color images were identified. These are flexible comparison in graphs, hidden Markov models, principal component analysis, and neural network methods. The analyzed methods of face recognition are mainly characterized by significant computational costs and low recognition performance. An exception is the neural network methods of face recognition, which, after completing the training, make it possible to obtain a high recognition performance at low computational costs. However, when changing the prototype images of faces, it often becomes necessary to redefine the network architecture and retrain the network. The specificity of neural network methods is also the complexity of selecting the network architecture and its training. Such papers are devoted to the use of neural networks onlyfor extraction of feature vectors of face images. The classification of the obtained feature vectors is then performed by knownmethods, namely, thresholding, a linear support vector machine, nearest neighbors, random forest. It has been observed that the lighting conditions in which the images were obtained and the turning of the head affect the shape of the separating surface and can decrease the feature vector classification performance for face images. Therefore, to improve the classification performance, it was decided to use correlation for prototype matching, a non-linear support vector machine and logistic regression. The performed experiment showed that correlation for prototype matching of low-light face images is characterized by higher classification performance compared to the thresholding. Moreover, the use of the Pearson and Spearman correlation coefficients showed similar results, and when using the Kendall correlation coefficient, the worst classification performance was obtained compared to the Pearson andSpearman coefficients. The research of the classification performance of images of faces that differ in head turn using correlation for prototype matching, a non-linear support vector machine and logistic regression showed the following. Correlation for prototype matching is more appropriate to use with small amounts of data due to the high classification performance and low computational complexity, since a small amount of data does not require a significant number of comparisons. However, on large amounts of data, the non-linear support vector machine requires less computation and shows similar classification performance. Using the results of the experiment, the researcher can select classification methods for a specific set of face images, preliminarily representing them with feature vectors using the network FaceNet.

Transferability analysis of adversarial attacks on gender classification to face recognition: Fixed and variable attack perturbation

Most deep learning-based image classification models are vulnerable to adversarial attacks that introduce imperceptible changes to the input images for the purpose of model misclassification. It has been demonstrated that these attacks, targeting a specific model, are transferable among models performing the same task. However, models performing different tasks but sharing the same input space and model architecture were never considered in the transferability scenarios presented in the literature. In this paper, this phenomenon was analysed in the context of VGG16-based and ResNet50-based biometric classifiers. The authors investigate the impact of two white-box attacks on a gender classifier and contrast a defence method as a countermeasure. Then, using adversarial images generated by the attacks, a pre-trained face recognition classifier is attacked in a black-box fashion. Two verification comparison settings are employed, in which images perturbed with the same and different magnitude of the perturbation are compared. The authors’ results indicate transferability in the fixed perturbation setting for a Fast Gradient Sign Method attack and non-transferability in a pixel-guided denoiser attack setting. The interpretation of this non-transferability can support the use of fast and train-free adversarial attacks targeting soft biometric classifiers as means to achieve soft biometric privacy protection while maintaining facial identity as utility.

Face recognition algorithm based on stack denoising and self-encoding LBP

Abstract To optimize the weak robustness of traditional face recognition algorithms, the classification accuracy rate is not high, the operation speed is slower, so a face recognition algorithm based on local binary pattern (LBP) and stacked autoencoder (AE) is proposed. The advantage of LBP texture structure feature of the face image as the initial feature of sparse autoencoder (SAE) learning, use the unified mode LBP operator to extract the histogram of the blocked face image, connect to form the LBP features of the entire image. It is used as input of the stacked AE, feature extraction is done, realize the recognition and classification of face images. Experimental results show that the recognition rate of the algorithm LBP-SAE on the Yale database has achieved 99.05%, and it further shows that the algorithm has a higher recognition rate than the classic face recognition algorithm; it has strong robustness to light changes. Experimental results on the Olivetti Research Laboratory library shows that the developed method is more robust to light changes and has better recognition effects compared to traditional face recognition algorithms and standard stack AEs.

Research on Face Recognition Classification Based on Improved GoogleNet

Face recognition is a relatively mature technology, which has some applications in many aspects, and now there are many networks studying it, which has indeed brought a lot of convenience to mankind in all aspects. This paper proposes a new face recognition technology. First, a new GoogLeNet-M network is proposed, which improves network performance on the basis of streamlining the network. Secondly, regularization and migration learning methods are added to improve accuracy. The experimental results show that the GoogLeNet-M network with regularization using migration learning technology has the best performance, with a recall rate of 0.97 and an accuracy of 0.98. Finally, it is concluded that the performance of the GoogLeNet-M network is better than other networks on the dataset, and the migration learning method and regularization help to improve the network performance.

STUDY AND ANALYSIS OF CLASSIFIERS FOR USE IN EMOTION RECOGNITION

Today, in the process of obtaining education, and not only, there is a widespread use of the method of communication through online platforms. The massive use of online platforms for communication creates the need for human emotion recognition systems. In facial recognition systems and human emotions, information is collected, analyzed and processed using special methods and learning algorithms. The paper considers and analyzes classifiers for face recognition, such as the HAAR classifier, the feature extraction method, local binary histogram (LBP), neural convolutional networks (CNN), which contribute to the effective implementation of emotion recognition methods.

A Survey on Classical and Modern Face Recognition Techniques

During the previous few centuries, facial recognition systems have become a popular research topic. On account of its extraordinary success and vast social applications; it has attracted significant study attention from a wide range of disciplines in the last five years - including “computer-vision”, “artificial-intelligence”, and “machine-learning”. As with most face recognition systems, the fundamental goal involves recognizing a person's identity by means of images, video, data streams, and context information. As a result of our research; we've outlined some of the most important applications, difficulties, and trends in scientific and social domains. This research, the primary goal is to summarize modern facial recognition algorithms and to gain a general perceptive of how these techniques act on diverse datasets. Aside from that, we also explore some significant problems like illumination variation, position, aging, occlusion, cosmetics, scale, and background are some of the primary challenges we examine. In addition to traditional face recognition approaches, the most recent research topics such as sparse models, deep learning, and fuzzy set theory are examined in depth. There's also a quick discussion of basic techniques, as well as a more in-depth. As a final point, this research explores the future of facial recognition technologies and their possible importance in the emerging digital society.

PCA-based dimensionality reduction for face recognition

In this paper, we conduct a comprehensive study on dimensionality reduction (DR) techniques and discuss the mostly used statistical DR technique called principal component analysis (PCA) in detail with a view to addressing the classical face recognition problem. Therefore, we, more devotedly, propose a solution to either a typical face or individual face recognition based on the principal components, which are constructed using PCA on the face images. We simulate the proposed solution with several training and test sets of manually captured face images and also with the popular Olivetti Research Laboratory (ORL) and Yale face databases. The performance measure of the proposed face recognizer signifies its superiority.

Comparison between Fisher’s Ratio and Information Gain with SVM classifier for 3 levels of enthusiasm classification through face recognition

The enthusiasm level of a person is an important measurement in real-world problems. This paper, therefore, presents face recognition classification for enthusiasm level, based on supervised machine learning, using Support Vector Machine (SVM) as a classifier, with a one-vs-one method because the data consists of more than two classes. In addition, Fisher’s Ratio and Information Gain are applied in the selection of contributive features, and the goals were to present an accuracy comparison between SVM and Fisher’s Ratio, as wells as with Information Gain, and the results showed the accuracy at 88,89%, and 80,95238%, respectively. This indicates the combination of SVM with Fisher Ratio to be better.