Age and Gender Recognition for Masked Face Using YOLO-X and CNN in Smart Advertisement Systems

Alexithymia, a multifaceted personality construct, is known to be related to difficulties in the decoding of emotional facial expressions, especially in case of suboptimal stimuli. The present study investigated whether and which facets of alexithymia are related to impairments in the recognition of emotions in faces with face masks. Accuracy and speed of emotion recognition were examined in a block of faces with and a block of faces without face masks in a sample of 102 healthy individuals. The order of blocks varied between participants. Emotions were recognized better and faster in unmasked than in masked faces. Recognition performance was worst and slowest for participants starting the task with masked faces. In the whole sample, there were no correlations of alexithymia facets with accuracy and speed of emotion recognition for masked and unmasked faces. In participants starting the task with masked faces, the facet externally oriented thinking was positively correlated with reaction latencies of correct responses for masked faces. Our findings indicate that an externally oriented thinking style could be linked to a less efficient identification of emotions from faces wearing masks when task difficulty is high and support the utility of a facet approach in alexithymia research.

Face recognition is a biometric technique that has been widely used in many fields. Most of the face recognition applications are used in the security and surveillance system. However, due to current pandemic, wearing facemasks is an obligation for everybody in public places. Hence, face recognition will encounter a problem such as extracting the facial features due to blockage caused by the facemasks. Thus, it will lower down the recognition rate level. The collected dataset consists of two categories which is masked face and unmasked face. This dataset called FaceMask Dataset was obtained from kaggle website. The Multi-Task Cascaded Neural Network (MTCNN) was used to find the face region in the dataset, and it will undergo image feature extraction and remove the undetected face as to prepare a proper training dataset before it can be trained by using LeNet algorithm. As the result, the categories in this work are fall into two classes, which masked face and unmasked face. The training accuracy was 98.21% and it will focus on the features on each result and will justify the difference between the accuracy for both classes. The proposed method was able to achieve 98.21% accuracy with LeNet algorithm as the face image was mainly focused on the face area without taking full size of the image.

Although effective in reducing virus transmission, face masks might compromise face recognition and trait judgments. With this study, we aimed to observe the influence of masks on face recognition and trait judgments—more specifically, in trustworthiness, dominance, and distinctiveness judgments. Also, we wanted to observe the possible influence of trait judgments on facial recognition for masked and unmasked faces, which has never been done before. For that, we conducted an online study where 140 participants observed and made trait judgments of masked and unmasked faces in a within-subjects design. After a distractive task, participants performed a recognition memory test. As expected, we observed a better recognition of faces shown without a mask during the study phase, which allowed the holistic processing of the faces. The worst performance was found for faces encoded with a mask but tested without it, occurring simultaneity disruption in holistic face processing and the violation of the encoding specificity principle. Regarding the trait judgments, unmasked faces were considered more distinctive, and masked faces were considered more trustworthy. More interestingly, we can conclude that facial distinctiveness predicts face recognition, regardless of mask use. In contrast, dominance judgments only predicted face recognition when faces were presented without a mask. When faces were exposed with masks, trustworthiness overrides dominance, becoming more critical to recognizing faces. We can interpret these results from an evolutionary perspective.

Dual-Proxy Modeling for Masked Face Recognition

In the current epidemic, the whole world is suffering with the infectious disease i.e., Corona Virus Disease (COVID-19). It is important to wear a mask to minimize the transmission of the disease. When everyone is wearing a face mask, it is difficult for recognition systems to recognize the masked face of a specific person. As some of the facial features are covered behind the mask e.g., mouth and nose. Therefore, the face-recognizing systems are inefficient to recognize the masked faces. To solve this issue, a face recognition system is proposed to recognize masked and unmasked faces. Support vector machine (SVM) and Random Forest (RF) based classifiers are trained on the specific dataset and classifiers effectively recognize the masked and unmasked faces. The classifier recognizes the human facial features such as eyes, eyebrows, forehead, ears, and hair. The dataset is collected in the form of images for 28 classes with and without a face mask. The trained system will recognize the person, whether the person is wearing a mask or not. The recognition accuracy is approximately 98.2% for different classes and the proposed recognizer is also compared with the state of art existing techniques.

Face recognition now requires a large number of labelled masked face images in the era of this unprecedented COVID19 pandemic. Unfortunately, the rapid spread of the virus has left us little time to prepare for such dataset in the wild. To circumvent this issue, we present a 3D model-based approach called WearMask3D for augmenting face images of various poses to the masked face counterparts. Our method proceeds by first fitting a 3D morphable model on the input image, second overlaying the mask surface onto the face model and warping the respective mask texture, and last projecting the 3D mask back to 2D. The mask texture is adapted based on the brightness and resolution of the input image. By working in 3D, our method can produce more natural masked faces of diverse poses from a single mask texture. To compare precisely between different augmentation approaches, we have constructed a dataset comprising masked and unmasked faces with labels called MFW-mini. Experimental results demonstrate WearMask3D <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">1</sup> produces more realistic masked faces, and utilizing these images for training leads to state-of-the-art recognition accuracy for masked faces.

To prevent the spread of COVID-19, face masks were mandatory in many public spaces around the world. Since faces are the gateway to early social cognition, this raised major concerns about the effect face masks may have on infants' attention to faces as well as on their language and social development. The goal of the present study was to assess how face masks modulate infants' attention to faces over the course of the first year of life. We measured 3, 6, 9, and 12-month-olds' looking behavior using a paired visual preference paradigm under two experimental conditions. First, we tested infants' preference for upright masked or unmasked faces of the same female individual. We found that regardless of age, infants looked equally long at the masked and unmasked faces. Second, we compared infants' attention to an upright masked versus an inverted masked face. Three- and 6-month-olds looked equally long to the masked faces when they were upright or inverted. However, 9- and 12-month-old infants showed a novelty preference for the inverted masked face. Our findings suggest that more experience with faces, including masked faces, leads to efficient adaptations of infants' visual system for processing impoverished social stimuli, such as partially occluded faces.

Wearing a facial mask has become a must in our daily life due to the global COVID-19 pandemic. However, the performance of conventional face recognition systems severely degrades for faces occluded by masks. How to combat the effect of occlusion on face recognition is an important issue. However, the performance of existing methods developed for masked face recognition unpleasantly degrades when dealing with unmasked faces. To address this issue for real-world applications, where the gallery image or the probe image may be a masked or unmasked face, we propose the concept of balanced facial feature matching and, based on it, design a robust masked face recognition system. The matching is balanced because it is performed on features extracted from corresponding facial regions. The system consists of a classification network and two feature extractors. The classification network classifies an input face image into a masked face or an unmasked face. One feature extractor extracts the feature of a full face, and the other uses a guided perceptual loss to focus the feature extraction on the non-occluded part of the face. The system is tested on both synthetic and real data. The face verification accuracy is improved by 2.4% for the synthetically masked LFW dataset, 1.9% for the MFR2 dataset, and 5.4% for the RMFD dataset. The results further show that the system improves masked face recognition while preserving the performance of unmasked face recognition.

COVID-19 is a life-threatening virus which affected people at a global level in just a matter of few months and is highly contagious. In order to reduce its spread, SOPs must be followed, such as washing hands, wearing face masks, and maintaining social distance. Hence, to aid the strict follow up of SOPs, this paper proposes a system to detect whether the people are wearing face masks and maintaining social distance or not in order to break the chain of COVID 19. The proposed system uses Deep Learning (DL) model based on Convolutional Neural Network (CNN) architecture for training the facemask detector and OpenPose 2D skeleton extraction technique for detecting social distance. A DL model based on a 7-layered CNN architecture was proposed in this research to detect masked and unmasked faces. Based on the proposed technique, 99.98% validation and 99.98% testing accuracies were achieved. In addition to that, the maintenance of social distance which is the new normal nowadays was also detected using the images obtained from the internet as currently, there is no such database available for detecting social distancing.

Deep-learning-based image inpainting methods have made remarkable advancements, particularly in object removal tasks. The removal of face masks has gained significant attention, especially in the wake of the COVID-19 pandemic, and while numerous methods have successfully addressed the removal of small objects, removing large and complex masks from faces remains demanding. This paper presents a novel two-stage network for unmasking faces considering the intricate facial features typically concealed by masks, such as noses, mouths, and chins. Additionally, the scarcity of paired datasets comprising masked and unmasked face images poses an additional challenge. In the first stage of our proposed model, we employ an autoencoder-based network for binary segmentation of the face mask. Subsequently, in the second stage, we introduce a generative adversarial network (GAN)-based network enhanced with attention and Masked-Unmasked Region Fusion (MURF) mechanisms to focus on the masked region. Our network generates realistic and accurate unmasked faces that resemble the original faces. We train our model on paired unmasked and masked face images sourced from CelebA, a large public dataset, and evaluate its performance on multi-scale masked faces. The experimental results illustrate that the proposed method surpasses the current state-of-the-art techniques in both qualitative and quantitative metrics. It achieves a Peak Signal-to-Noise Ratio (PSNR) improvement of 4.18 dB over the second-best method, with the PSNR reaching 30.96. Additionally, it exhibits a 1% increase in the Structural Similarity Index Measure (SSIM), achieving a value of 0.95.

Protective face masks were one of the central measures to counteract viral transmission in the COVID-19 pandemic. Prior research indicates that face masks impact various aspects of social cognition, such as emotion recognition and social evaluation. Whether protective masks also influence social avoidance behavior is less clear. Our project assessed direct and indirect measures of social avoidance tendencies towards masked and unmasked faces in two experiments with 311 participants during the first half of 2021. Two interventions were used in half of the participants from each sample (Experiment 1: protective face masks; Experiment 2: a disease prime video) to decrease or increase the salience of the immediate contagion threat. In the direct social avoidance measure, which asked for the deliberate decision to approach or avoid a person in a hypothetical social encounter, participants showed an increased willingness to approach masked as opposed to unmasked faces across experiments. This effect was further related to interindividual differences in pandemic threat perception in both samples. In the indirect measure, which assessed automatic social approach and avoidance tendencies, we neither observed an approach advantage towards masked faces nor an avoidance advantage for unmasked faces. Thus, while the absence of protective face masks may have led to increased deliberate social avoidance during the pandemic, no such effect was observed on automatic regulation of behavior, thus indicating the relative robustness of this latter behavior against changes in superordinate social norms.

Previous studies in the context of the COVID-19 pandemic indicated that wearing a medical-style mask affects whether a stranger’s face is judged as more trustworthy, socially desirable, or likely to be ill. However, given political controversies around mask use, these effects might vary by political orientation. In a pre-registered online experiment, we measured evaluations of trustworthiness, social desirability and perceived illness in masked and unmasked faces by 1241 British and US participants. We included questions on political orientation, along with the implicit online-VAAST approach/avoid task to test reaction times to masked/unmasked faces. There was a medium-sized effect of masks on trustworthiness and a significant interaction with political orientation, in that conservatives found masked faces less trustworthy than did liberals. Participants were quicker to approach masked than unmasked faces, but conservatives were relatively slower than liberals. The effects on trustworthiness suggest that differential moralization of novel social norms can affect how their adherents are evaluated in terms of their suitability for social interactions. Furthermore, the congruence between implicit and explicit methods implies that such differences can have deep-seated effects on reactions.

The unprecedented efforts to minimize the effects of the COVID-19 pandemic introduce a new arena for human face recognition in which faces are partially occluded with masks. Here, we tested the extent to which face masks change the way faces are perceived. To this end, we evaluated face processing abilities for masked and unmasked faces in a large online sample of adult observers (n = 496) using an adapted version of the Cambridge Face Memory Test, a validated measure of face perception abilities in humans. As expected, a substantial decrease in performance was found for masked faces. Importantly, the inclusion of masks also led to a qualitative change in the way masked faces are perceived. In particular, holistic processing, the hallmark of face perception, was disrupted for faces with masks, as suggested by a reduced inversion effect. Similar changes were found whether masks were included during the study or the test phases of the experiment. Together, we provide novel evidence for quantitative and qualitative alterations in the processing of masked faces that could have significant effects on daily activities and social interactions.

This paper demonstrates the usability of thermal video for facial mask detection and the breathing rate measurement. Due to the lack of available thermal masked face images, we developed a dataset based on the SpeakingFaces set, by generating masks for the unmasked thermal images of faces. We utilize the Cascade R-CNN as the thermal facial mask detector, identifying masked and unmasked faces, and whether the mask colour indicates a inhale or exhale state. The latter is used to calculate the breathing rate. The proposed Cascade R-CNN is a multi-stage object detection architecture composed of detectors trained with increasing Intersection-of-Unions thresholds. In our experiments on the Thermal-Mask dataset, the Cascade R-CNN achieves 99.7% in precision, on average, for the masked face detection, and 91.1% for recall. To validate our approach, we also recorded a small set of videos with masked faces to measure the breathing rate. The accuracy result of 91.95% showed a promising advance in identifying possible breath abnormalities using thermal videos, which may be useful in screening subject for COVID-19 symptoms.

Use of face masks is one of the measures adopted by the general community to stop the transmission of disease during this ongoing COVID-19 pandemic. This wide use of face masks has indeed been shown to disrupt day-to-day face recognition. People with autism spectrum disorder (ASD) often have predisposed impairment in face recognition and are expected to be more vulnerable to this disruption in face recognition. Here, we recruited typically developing adult participants and those with ASD, and we measured their non-verbal intelligence, autism spectrum quotient, empathy quotient, and recognition performances of faces with and without a face mask covering the lower halves of the face. When faces were initially learned unobstructed, we showed that participants had a general reduced face recognition performance for masked faces. In contrast, when masked faces were first learned, typically developing adults benefit with an overall advantage in recognizing both masked and unmasked faces; while adults with ASD recognized unmasked faces with a significantly more reduced level of performance than masked faces—this face recognition discrepancy is predicted by a higher level of autistic traits. This paper also discusses how autistic traits influence processing of faces with and without face masks.