Face Recognition Applications Research Articles

Face recognition with masks in the lasting era of COVID-19

At the end of 2019, the pandemic began to have a huge impact on people's lives around the world. In order to control and prevent the virus from spreading, world-wide governments have encouraged all residents to wear masks in their daily lives. Since the 1960s, face recognition technology has gradually emerged and developed. In the 21st century, it has been widely used in security, finance, transportation, and even retail, advertising, smart devices, education, health care, entertainment and other fields, playing a pivotal role in everyone's life. In the era of COVID-19, the public wearing face masks has undoubtedly hindered the smooth technology application of face recognition. The technology of recognizing human face in the case of wearing masks should be developed, followed by the adjustment of the original technology and data. The development of technology also brings certain social impact. Our research reviews the adjustment and progress of face recognition technology in the era of COVID-19, summarizes the impact of technological progress on society and people's lives, and provides help for the government to better prevent and control the epidemic.

Analysis of the application of facial expression recognition in psychotherapy

Peoples psychological problems are increasing with the increasing influence of COVID-19. Psychological problems can greatly affect our daily lives, so we must take this problem seriously. Traditional psychotherapy has many disadvantages. To address these shortcomings, a new technology that combines artificial intelligence is gradually emerging in the psychotherapy industry. Use image recognition in artificial intelligence to help consultants do this job better. This article discusses how technologies in artificial intelligence such as facial expression recognition, chatbots, and other technologies can improve these shortcomings and presents hypotheses about the future of psychotherapy and artificial intelligence based on relevant The results show that although facial expression recognition has broad prospects, there are still many problems, such as fewer micro expression data sets and insufficient training models for model generalization ability.

Design System Using Matlab to Recognize Patterns as Face Detection

This paper presents the design and implementation of a face detection system using MATLAB. The system utilizes various techniques, including preprocessing, feature extraction, and pattern recognition algorithms, to accurately detect faces in images. The methodology section provides a detailed explanation of the system's workflow, highlighting the preprocessing techniques for enhancing image quality, the feature extraction methods for capturing discriminative facial characteristics, and the utilization of SVM-based algorithms for pattern recognition. The system's performance is evaluated using established metrics such as accuracy, precision, recall, and F1 score, and comparisons with existing approaches are made. The experimental setup, including data partitioning, parameter optimization, and computational efficiency analysis, is described to ensure reproducibility and reliability. The results demonstrate the system's effectiveness in accurately detecting faces in various scenarios. Future work can focus on further enhancing the system's performance and expanding its applications in face recognition and facial expression analysis.

Evaluating Trustworthiness and Racial Bias in Face Recognition APIs Using Deepfakes

We evaluate the racial bias in face recognition application programming interfaces (APIs) using real and deepfake celebrity images. We use deepfake generation methods to introduce small, imperceptible changes to the real images to shift the racial class of predictions, showing how deepfake images exacerbated racial bias in web-based face recognition APIs.

A simple and effective patch-Based method for frame-level face anti-spoofing

With the wide applications of face recognition, face anti-spoofing has become a major challenge for reliable face recognition. Thus, it is necessary to perform face liveness detection. Most existing methods rely on a whole face image for training and testing and are thus susceptible to the overfitting problem because of limited training samples; meanwhile, liveness information is not fully explored. To address these issues, we propose a simple and effective patch-based approach. There are two main contributions: 1) different patch sampling strategies are applied to a training set and a testing set to overcome the overfitting problem, and 2) an attention mechanism is applied to explore more significant information for liveness detection. We evaluate the proposed approach on four popular and challenging databases: the CASIA-SURF, OULU-NPU, CASIA-FASD and REPLAY-ATTACK databases. The proposed method could obtain very promising liveness detection performance. For example, the average classification error rate (ACER) on the CASIA-SURF database (using RGB images only) was 1.6%, which is the lowest reported error rate to the best of our knowledge.

FERFM: An Enhanced Facial Emotion Recognition System Using Fine-tuned MobileNetV2 Architecture

Deep learning techniques have recently drawn a lot of attention towards automatic facial emotion recognition (FER) applications such as diagnosis of psychiatric disorders, mental health monitoring, and other human–computer interaction applications. But the models that have been developed, have still not been generalized due to the absence of substantial emotion datasets and high operational costs. To address these issues, we propose a novel technique called facial emotion recognition using fine-tuned MobileNetV2 architecture (FERFM). It uses the concept of transfer learning to improve the performance of FER systems for mobile devices. It transfers the knowledge from pre-trained ImageNet dataset to profile view RGB-KDEF dataset for diversity among learnt features. Initially, data descriptive analysis is performed on input KDEF images to automate the necessary knowledge. Then the pipeline strategy is introduced, where the pre-trained MobileNetV2 architecture is fine tuned by eliminating the last six layers and adding dropout regularization, max pooling and dense layer. To test the performance of the proposed work, we ran several tests using the KDEF-RGB dataset, FER13 dataset and real-time facial expression images to classify the results into seven discrete categories of emotions. It was also verified on VGG-16 pretrained model and demonstrated remarkable performance comparatively. Proposed FERFM achieves 85.7% accuracy and requires 1,510,599 number of trainable parameters, 43 ms running time per image that proves the enhancement of FER performance and outperforms the existing state of art methods. Also, the ability to achieve low operational costs while maintaining high accuracy demonstrates proficiency in mobile applications.

Application of facial expression recognition based on domain-adapted convolutional neural network in English smart teaching system

Application of facial expression recognition based on domain-adapted convolutional neural network in English smart teaching system

Masked Face Recognition Dataset and Application

During COVID-19 coronavirus epidemic, almost everyone wears a mask to prevent the spread of virus. It raises a problem that the traditional face recognition model basically fails in the scene of face-based identity verification, such as security check, community visit check-in, etc. Therefore, it is imminent to boost the performance of masked face recognition. Most recent advanced face recognition methods are based on deep learning, which heavily depends on a large number of training samples. However, there are presently no publicly available masked face recognition datasets, especially real ones. To this end, this work proposes three types of masked face datasets, including Masked Face Detection Dataset (MFDD), Real-world Masked Face Recognition Dataset (RMFRD) and Synthetic Masked Face Recognition Dataset (SMFRD). Besides, we conduct benchmark experiments on these three datasets for reference. As far as we know, we are the first to publicly release large-scale masked face recognition datasets that can be downloaded for free at https://github.com/X-zhangyang/Real-World-Masked-Face-Dataset..

Face Recognition Technique using YOLO V5 Algorithm

In today’s date real world application like human-machine interaction, security surveillance face recognition has made its great importance. For face recognition the steps to be followed are data collection, preprocessing, Feature Extraction, Training Evaluation and finally testing. One of the best Algorithms used for face recognition is Viola-Jones Algorithm. Viola Jones Algorithm is highly accepted because of its fast processing time and high detection rate. The other detection Algorithms which can be used are HOG Algorithm (Histogram Oriented Gradient), Deep Learning CNN (Convolution Neural Network) Algorithm, Haar cascade Algorithm and MTCNN Algorithm. In this Paper we will discuss about all this Algorithms and compare the detection rate of facial points among various methods for the datasets. YOLO V5 algorithm is another algorithm which can be used for face recognition technique. Face recognition can be achieved by combining the YOLO V5 algorithm with additional technique such as deep face recognition models like face net, which can recognize and compare facial features to identify individuals.YOLOV5 can identify facial features such as eyes, mouth, and nose, making it useful for a variety of face recognition applications.

HYBRID END-TO-END APPROACH INTEGRATING ONLINE LEARNING WITH FACE-IDENTIFICATION SYSTEM

To date, facial recognition has been one of the most intriguing, interesting research topics over years. It requires some specific face-based algorithms such as facial detection, facial alignment, facial representation, and facial recognition as well; however, all of these algorithms derive from heavy deep learning architectures that cause limitations for development, scalability, flawed accuracy, and deployment into publicity with mere CPU servers. It also calls for large datasets containing hundreds of thousands of records for training purposes. In this paper, we propose a full pipeline for an effective face recognition application which only uses a small Vietnamese celebrity dataset and CPU for training that can solve the leakage of data and the need for GPU devices. It is based on a face vector-to-string tokens algorithm then saves face’s properties into Elasticsearch for future retrieval, so the problem of online learning in Facial Recognition is also tackled. Comparison with another popular algorithm on the dataset, our proposed pipeline not only outweighs the accuracy counterpart, but it also achieves a very speedy time inference for a real-time face recognition application.

A low-resolution real-time face recognition using extreme learning machine and its variants

ABSTRACT Extreme Learning Machine (ELM) is an excellent candidate for its exemplary performance as a generalized Single Layer Feed-forward Network (SLFN). In this regard, ELM has attracted widespread attention for tackling multiclass classification and regression problems with relative ease. Due to the random allocation of weights to the input and hidden layer biases, the running time spans of ELM are observed to be in seconds and agree with real-time events whilst giving a competitive performance. In this work, the suitability of ELM and its Online Sequential variant for recognition of unknown face samples is investigated under uncontrolled environments, including varying pose, variance, and illumination. The presented face recognition approach is evaluated sing four datasets from YALE, CMU, BIOID, and LFW. To track the face features of a sample face in its spatial domain with varying windows corresponding to different datasets, the Viola-Jones object detection framework is used. The face features are then extracted and evaluated using a Histogram of Oriented Gradients, which forms the dataset to be fed to ELM and OS-ELM classifiers. The running time for training the proposed face recognition approach is found to span within milliseconds to seconds with the computed time complexity of , justifying the suitability of ELM and OS-ELM classifiers for real-time face recognition applications with a high recognition rate outcome.

On-road driver facial expression emotion recognition with parallel multi-verse optimizer (PMVO) and optical flow reconstruction for partial occlusion in internet of things (IoT)

In recent years, many large-scale information systems in the Internet of Things (IoT) can be converted into interdependent sensor networks, such as smart cities, smart medical systems, and industrial Internet systems. The successful application of Facial Expression Recognition in the IoT will make our algorithms faster, more convenient, lower overall costs, providing better business practices, and enhance sustainability. Facial Expression Recognition (FER)is essential to effectively communicate between human and machines. Video facial expression detection is a crucial component for gauging the driver's mood for driver assistance system. The emotions of the driver have a significant role in dictating the behavior of the driver, according to several studies, which can lead to disastrous car crashes. However, criteria affecting the identification of driver emotions with the right kind of monitoring include changes in stance, lighting, and occlusions. Based on the restoration of the blurred facial region, the Driver Facial Expression Emotion Recognition (DFEER) system was developed to address these issues. One of the first things to do when dealing with a sequence of blurred faces is to calculate the optical fluxes between the frames. After that, an optical flow reconstruction using a trained PMVO is performed to fix the occlusion-induced damage (Parallel Multi-Verse Optimizer). The major solutions are randomly split into groups of occluded and non-occluded optical flows using a parallel technique, and the groups discuss their findings after a set number of iterations. After optical flows have been rebuilt, they are used immediately in the classification phase of expression prediction. In this study, Very Deep Convolution Networks (VGGNet) proposed a method for recognising human emotions. Both the CK+ (Cohn-Kanade database) and the KMU-FED (Keimyung University Facial Expression of Drivers) databases are used to carry out the assessments of the classification model's efficacy. Accuracy, recall, precision, and f-measure the effectiveness of the suggested strategy is then evaluated using the findings.

Fairness-aware training of face attribute classifiers via adversarial robustness

Developing fair deep learning models for identity-sensitive applications (e.g., face attribute recognition) has gained increasing attention from the research community. Indeed, it has been observed that deep models can easily overfit to the bias of the training set, resulting in discriminative performance against certain demographic groups during test time. Motivated by the observation that a biased classifier could result in different adversarial robustness among training samples in different demographic groups (robustness bias), we argue that such adversarial robustness information of individual training samples could imply whether the training data distribution is fair among different demographic groups. In other words, under a fair classifier, the training samples from different demographic groups are expected to show similar or comparable adversarial robustness. Therefore, in this work, we propose to re-weight the training loss of individual training samples using their adversarial robustness, and provide the fairness-awareness in the training process. Extensive experimental results on CelebA dataset show that the face attribute classifiers could learn significantly greater demographic fairness under our proposed training objective and outperform other state-of-the-art re-weighting fairness algorithms on different face recognition applications. Moreover, our proposed method also reduces the non-trivial robustness bias among different demographic groups, preventing the under-represented demographic groups from higher adversarial threats.

New body perimeter and biometrics as personal data

This article deals with the new body perimeter and biometrics as personal data, seeking, from a principled point of view, to present recommendations for the application of facial recognition in the São Paulo Metro. Balancing the impacts of biometric techniques and the principles of purpose and necessity of the Brazilian General Data Protection Law, the following question arises: can facial biometrics be applied on the premises of the São Paulo Metro? And, if the answer is positive, what would be its limits, based on the aforementioned principles? With regard to methodology, the research was theoretical, dealing with the theme in an exploratory and descriptive way, making use of bibliographic technical procedures.

Face comparison analysis of patients with orthognathic surgery treatment using cloud computing–based face recognition application programming interfaces

This study aimed to investigate whether the postoperative change in patients after orthognathic surgery, whose facial aesthetics was affected, led to detectable differences using Microsoft Azure, Amazon Web Services Rekognition, and Face++, which were commercially available face recognition systems. Photographs of 35 patients after orthognathic surgery were analyzed using 3 well-known cloud computing-based facial recognition application programming interfaces to compute similarity scores between preoperative and postoperative photographs. The preoperative, relaxed, smiling, profile, and semiprofile photographs of the patients were compared separately to validate the relevant application programming interfaces. Patient characteristics and type of surgery were recorded for statistical analysis. Kruskal-Wallis rank sum tests were performed to analyze the relationship between patient characteristics and similarity scores. Multiple-comparison Wilcoxon rank sum tests were performed on the statistically significant characteristics. The similarity scores in the Face++ program were lower than those in the Microsoft Azure and Amazon Web Services Rekognition. In addition, the similarity scores were higher in smiling photographs. A statistically significant difference was found in similarity scores between relaxed and smiling photographs according to different programs (P<0.05). For all 3 facial recognition programs, comparable similarity scores were found in all photographs taken before and after surgery across sex, type of surgery, and type of surgical approach. The type of surgery and surgical approach, sex, and amount of surgical movement did not significantly affect similarity scores in any facial recognition programs (P>0.05). The similarity scores between the photographs before and after orthognathic surgery were high, suggesting that the software algorithms might value measurements on the basis of upper-face landmarks more than lower-face measurements.

CMOS-GAN: Semi-supervised Generative Adversarial Model for Cross-Modality Face Image Synthesis.

Cross-modality face image synthesis such as sketch-to-photo, NIR-to-RGB, and RGB-to-depth has wide applications in face recognition, face animation, and digital entertainment. Conventional cross-modality synthesis methods usually require paired training data, i.e., each subject has images of both modalities. However, paired data can be difficult to acquire, while unpaired data commonly exist. In this paper, we propose a novel semi-supervised cross-modality synthesis method (namely CMOS-GAN), which can leverage both paired and unpaired face images to learn a robust cross-modality synthesis model. Specifically, CMOS-GAN uses a generator of encoder-decoder architecture for new modality synthesis. We leverage pixel-wise loss, adversarial loss, classification loss, and face feature loss to exploit the information from both paired multi-modality face images and unpaired face images for model learning. In addition, since we expect the synthetic new modality can also be helpful for improving face recognition accuracy, we further use a modified triplet loss to retain the discriminative features of the subject in the synthetic modality. Experiments on three cross-modality face synthesis tasks (NIR-to-VIS, RGB-to-depth, and sketch-to-photo) show the effectiveness of the proposed approach compared with the state-of-the-art. In addition, we also collect a large-scale RGB-D dataset (VIPL-MumoFace-3K) for the RGB-to-depth synthesis task. We plan to open-source our code and VIPL-MumoFace-3K dataset to the community.

Exploring Deep Learning Ear Recognition in Thermal Images

Ear biometrics is an appealing choice for human recognition in various daily activities, ranging from surveillance to commercial-related applications. It has multiple advantages in identifying non-cooperative individuals in visible unconstrained environments. Moreover, Infrared imagery is an alternative sensing modality for ear recognition systems in the dark or when there is no control over illumination. In this work, we investigate the performance of deep learning models for ear recognition in the visible and thermal infrared domains. We examine multiple convolutional neural network architectures and develop different strategies to enhance learning. We use a dual-band dataset for our experiments, recently collected for multi-pose (full frontal to full profile) face recognition applications. We also use other ear datasets for learning, the AWE visible Ear Dataset & the WVU MIWR Profile Face Dataset, for visible domain and thermal domain learning, respectively. Our experiments achieve a 98.76% Rank-1 identification accuracy, 0.7596% Equal Error Rate (EER) verification for the visible domain ear recognition, and 96.93% Rank-1 identification accuracy, 2.4541% EER verification for the thermal domain ear recognition.

Comparing pre-trained algorithms for facial emotion recognition: An analysis

Facial emotion recognition is an evolving field that aims to detect and interpret human emotions based on facial expressions. This technology combines computer vision, machine learning, and artificial intelligence techniques to analyze facial features and accurately classify emotions such as happiness, sadness, anger, fear, surprise, and disgust. The human face is a powerful channel for emotional communication, and facial emotion recognition seeks to replicate this ability using computational algorithms. By capturing and analyzing facial cues such as eyebrow movements, eye widening, lip curvature, and changes in skin color, these systems can infer emotional states in real-time. The applications of facial emotion recognition are diverse and impactful. In psychology, it enables researchers to study emotional responses and behavior, providing insights into mental health conditions and aiding therapists in understanding their clients' emotional states. In human-computer interaction, this technology allows for more intuitive and personalized interactions, enhancing user experiences. In marketing and advertising, facial emotion recognition helps companies understand consumer emotional engagement and tailor their strategies accordingly. In healthcare, it can aid in the diagnosis and monitoring of mental health conditions, enabling personalized treatment options.

Data Mining Approach Based on Hierarchical Gaussian Mixture Representation Model

Infinite Gaussian mixture process is a model that computes the Gaussian mixture parameters with order. This process is a probability density distribution with adequate training data that can converge to the input density curve. In this paper, we propose a data mining model namely Beta hierarchical distribution that can solve axial data modeling. A novel hierarchical Two-Hyper-Parameter Poisson stochastic process is developed to solve grouped data modelling. The solution uses data mining techniques to link datum in groups by linking their components. The learning techniques are novel presentations of Gaussian modelling that use prior knowledge of the representation hyper-parameters and approximate them in a closed form. Experiments are performed on axial data modeling of Arabic Script classification and depict the effectiveness of the proposed method using a hand written benchmark dataset which contains complex handwritten Arabic patterns. Experiments are also performed on the application of facial expression recognition and prove the accuracy of the proposed method using a benchmark dataset which contains eight different facial expressions.

Using Open-Source Image Datasets for Research

The application of AI-based Facial Recognition (FRT) raises various legal and ethical questions. A comprehensive analysis must take into consideration not only the creation and future use of facial recognition but has to start even earlier: namely with the selection of the underlying data sets. While it may seem practical to utilize open-source image datasets for the creation, training and testing of models, the compliant use of such data sets still represents a major legal hurdle, especially with regard to international aspects. This is mainly due the fact, that harmonization in the area of research and data protection is very limited and still largely left to national legislators, even within the European Union. In this article, the authors analyse key problems of the use of open-source datasets for research in the area of FRT based on AI.