Single Face Research Articles

Monogenic features based single sample face recognition by kernel sparse representation on multiple Riemannian manifolds

This paper presents a novel method for single sample face recognition, characterized by the use of grayscale monogenic features of the face images, and the kernel sparse representation on multiple Riemannian manifolds. To indicate regional face discriminability, the multi-scale extended monogenic features are firstly locally extracted from different regions of an image, according to a specific face partition scheme, and then the intrinsic subspace of the feature vector set corresponding to each region is further extracted and modeled as a point of a Grassmann manifold. The congeneric local feature scheme is also exploited to entire image for the extraction of co-occurrence distributions of the grouped feature images of intersectional dimensions, based on a special binarization scheme applied to the resultant feature images. This derives the auxiliary marginal distribution-based descriptors residing on the closures of multiple multinomial manifolds. To train the kernel sparse representation classifier using the combined descriptors for a recognition task, the strategy of kernel alignment combining column L2-norm-based kernel matrix normalization are adopted for multiple kernel fusion, where the used kernels are all derived from Riemannian geometries of two types of manifolds. The superiority of our method is demonstrated through extensive experiments.

Physiological Effects of Single and Double Face Mask Use with Moderate and Vigorous Exercise

ABSTRACT Introduction/Purpose Exercise is associated with increased exhalation of infectious particles in respiratory disease, and face mask use has become routine during the coronavirus disease 2019 (COVID-19) pandemic to mitigate particle spread. As of February 2022, the U.S. Centers for Disease Control and Prevention (CDC) continues to recommend the use of double masking (i.e., cloth masks worn over a surgical mask) to minimize particle leakage seen with single-mask use. Studies to date have examined the effects of single masks on exercise safety and performance, but the potential added burden during exercise of an additional mask layer has not been evaluated. The purpose of our study was to compare the effects of single and double face mask use during moderate to vigorous exercise. Methods Fifteen healthy, physically active young adults (mean age = 21.1 ± 0.8 yr; 8 males and 7 females; V˙O2peak = 46.3 ± 11.5 mL·kg−1⋅min−1) completed the study. This included an initial assessment of V˙O2peak, with subjects returning for three exercise test sessions performed while wearing either a surgical face mask, a cloth face mask, or double masks (i.e., a cloth mask worn over a surgical mask) in a random order. Test sessions consisted of a progressive cycling protocol with 10-min stages at 40%, 60%, and 75% oxygen consumption reserve (VO2R). Assessments were performed after 5 and 10 min at each exercise intensity and included heart rate, respiratory rate, oxygen saturation, and ratings of perceived exertion and dyspnea. Results At exercise intensities from 40% to 75% VO2R, there were no significant differences between cloth masks, surgical masks, and double masking in the measured subjective and physiologic parameters. Conclusions When compared with surgical and cloth face masks in common use, for the outcomes measured in this study, double masking in accordance with CDC recommendations does not result in significant impairments during moderate to vigorous exercise up to 75% VO2R in healthy young adults.

Single Parents Competing in a Dual-Earner Society: Social Policy to Level the Playing Field

I examine the relative poverty risk among single-parent households in countries that have a large share of households with dual earners. Data from the Luxembourg Income Study (LIS) Database are used to analyze eighteen OECD countries in the period 1984 to 2010. I find that single parents face higher relative income poverty risks in countries with a large share of dual-earner households and that this higher risk of poverty is related to higher standards of living in those countries: higher standards of living have raised poverty thresholds, and single-parent incomes are less likely to reach those higher poverty thresholds. I also find that this overall pattern varied across institutional contexts: a rise of dual-earner households puts single parents at a disadvantage only in countries that have relatively low public expenditures on childcare and relatively low income transfer policies.

Multi-View Graph Fusion for Semi-Supervised Learning: Application to Image-Based Face Beauty Prediction

Facial Beauty Prediction (FBP) is an important visual recognition problem to evaluate the attractiveness of faces according to human perception. Most existing FBP methods are based on supervised solutions using geometric or deep features. Semi-supervised learning for FBP is an almost unexplored research area. In this work, we propose a graph-based semi-supervised method in which multiple graphs are constructed to find the appropriate graph representation of the face images (with and without scores). The proposed method combines both geometric and deep feature-based graphs to produce a high-level representation of face images instead of using a single face descriptor and also improves the discriminative ability of graph-based score propagation methods. In addition to the data graph, our proposed approach fuses an additional graph adaptively built on the predicted beauty values. Experimental results on the SCUTFBP-5500 facial beauty dataset demonstrate the superiority of the proposed algorithm compared to other state-of-the-art methods.

Evaluation of single shear performance of nailed joint damaged by cyclic deformation due to moderate earthquakes

A single face shear experiment with numerous small deformations of the nailed joint was performed on plywood, OSB, MDF, and gypsum board to confirm whether cyclic deformations caused by moderate earthquakes affect the shear resisting wall of a wooden house. As a result, it was found that the maximum load and yield load are not significantly affected even when repeatedly small deformed. The reduction ratio of the load of gypsum boards was higher than that of other boards due to repeated deformation. Additionally, the equivalent stiffness of repeated specimens was reduced accordingly. It was supposed that the equivalent stiffness of the shear resisting wall decreases when it is repeatedly deformed.

Study on irradiation effects of refractory bcc high-entropy alloy

• Refractory HEAs exhibit high thermal stabilities after heat-treatment. • The irradiation hardening of MoNbTaTiW is smaller than that of MoNbTaTi. • Refractory HEAs have higher irradiation resistance than pure tungsten. We developed body-centered cubic refractory high-entropy alloys, MoNbTaTi, MoNbTaW, and MoNbTaTiW, and explored their microstructure, and mechanical properties in order to develop a new high-irradiation resistance material suitable for fusion reactor components. Three HEAs were annealed at 1200 °C for 48 h to form single face BCC alloys, while MoNbTaW was also annealed at 1500 °C. SEM inspection and EDS mapping have both validated this result. The Vickers hardness of three HEAs was investigated, and the stable results indicate that they are all stable at elevated temperatures. Three HEAs are ion irradiated with 6.4 MeV Fe 3+ at a temperature of 500 °C, and the irradiation hardening is examined using nanoindentation. In comparison to pure tungsten, the three HEAs exhibit increased resistance to irradiation.

Age Estimation of Faces in Videos Using Head Pose Estimation and Convolutional Neural Networks.

Age estimation from human faces is an important yet challenging task in computer vision because of the large differences between physical age and apparent age. Due to the differences including races, genders, and other factors, the performance of a learning method for this task strongly depends on the training data. Although many inspiring works have focused on the age estimation of a single human face through deep learning, the existing methods still have lower performance when dealing with faces in videos because of the differences in head pose between frames, which can lead to greatly different results. In this paper, a combined system of age estimation and head pose estimation is proposed to improve the performance of age estimation from faces in videos. We use deep regression forests (DRFs) to estimate the age of facial images, while a multiloss convolutional neural network is also utilized to estimate the head pose. Accordingly, we estimate the age of faces only for head poses within a set degree threshold to enable value refinement. First, we divided the images in the Cross-Age Celebrity Dataset (CACD) and the Asian Face Age Dataset (AFAD) according to the estimated head pose degrees and generated separate age estimates for images with different poses. The experimental results showed that the accuracy of age estimation from frontal facial images was better than that for faces at different angles, thus demonstrating the effect of head pose on age estimation. Further experiments were conducted on several videos to estimate the age of the same person with his or her face at different angles, and the results show that our proposed combined system can provide more precise and reliable age estimates than a system without head pose estimation.

Cycle-autoencoder based block-sparse joint representation for single sample face recognition

Single sample face recognition (FR) uses only one image per person for training. It is a challenging task due to insufficient training samples and dramatic variance of unseen images in real applications. In this paper, we propose a cycle-autoencoder model to generate facial variation from the single training sample and remove the variation in the testing set. Our approach adopts a cycle consistency scheme to formulate the generation and removal models in one framework. Considering the prior structure of the images produced from the generation and removal models, we further propose a block-sparse joint representation method, which integrates the representation procedure of all testing samples and obtains all the coefficients simultaneously. The experimental results on AR, Extended Yale B and CUFS datasets not only demonstrate the effectiveness of our proposed method, but also represent its robustness to complex facial variations. Experiments on the CUFS dataset also show that our approach is appropriate for sketch face recognition with single sample per person.

DAP5 Binding to the FGF‐9 mRNA 5’ UTR is Structurally‐Dependent

Eukaryotic mRNA translation is typically modulated by eukaryotic initiation factor 4E (eIF4E) binding to the 5’‐m7G cap. However, under the hypoxic conditions within the cancer cell, eIF4E is segregated by 4E binding protein. Thus, cap‐independent recruitment of the 40S ribosome to the 5’ untranslated region (5’ UTR) of the mRNA is required for tumorigenesis. Fibroblast growth factor 9 (FGF‐9) is one such angiogenic growth factor upregulated in a cap‐independent manner in breast and colorectal cancer cells. Previously, our lab has demonstrated that binding of death‐associated factor 5 (DAP5), a eukaryotic initiation factor, to the FGF‐9 5’ UTR directly impacts cap‐independent translational efficiency. However, the exact binding site of DAP5 within the 5’ UTR is unknown. RNA secondary and tertiary structure within non‐coding regions has been demonstrated to play a role in cap‐independent translational efficiency; therefore, we hypothesized that specific structures act as eukaryotic initiation factor binding sites. Using SHAPE‐sequencing data, we reveal that the 5’ UTR of FGF‐9 is highly structured and possesses a single‐stranded region near the start codon. Through SHAPE footprinting and toeprinting experiments, we also demonstrate that DAP5 binds along a single face of the FGF‐9 5’ UTR and contacts the 3’ single‐stranded region. Our lab had previously reported that cap‐independent translation of FGF‐9 does not require a free 5’ end, suggesting little to no ribosomal scanning for the start codon occurs. Our current data demonstrate that initiation factor binding is structurally‐dependent and suggest that limited ribosomal scanning is due to DAP5‐dependent restructuring of the 5’ UTR to bring the start codon within striking distance of the 40S ribosome. We are currently examining whether cap‐independent upregulation of other tumorigenic transcripts follows a similar pattern.

Performance Comparison of Single and Double Masks: Filtration Efficiencies, Breathing Resistance and CO2 Content.

The emergence of a pandemic affecting the respiratory system has resulted in a significant demand for face masks. Masks have always been mentioned as an effective tool against environmental threats. This includes the use of double masks by large sections of the public, as can be seen during the current global spread of COVID-19. However, there is limited knowledge available on the performance of the various commonly available double face masks used. In this study, we have compared the performance of single and double face configurations. Three types of 3-ply face mask (FM), three types of fabric face mask (CM) and nine configurations of these face masks were investigated based on morphology, filtration efficiencies, breathing resistance and carbon dioxide (CO2) content. The filtration efficiencies of the double face masks are enhanced when combined with a fabric mask, and were found to improve by 500% compared to those of a fabric mask. The lowest breathing resistance among the double masking was observed in the combination of FM1 and fabric face masks. From the results, it was found that all combinations of double face mask are suitable to use since the measurements indicate that the breathing resistance and CO2 contents are still lower than the acceptable value set by the Malaysian Standard (MS 2323:2010) and European Standard (EN 149:2001 + A1:2009). The performance of double masks is comparable with that of N95 respirators. The double masking combination can be used as an alternative method to reduce inhalation exposure to airborne contaminants.

CNN-Based Restoration of a Single Face Image Degraded by Atmospheric Turbulence

Atmospheric turbulence significantly affects imaging systems which use light that has propagated through long atmospheric paths. Images captured under such condition suffer from a combination of geometric deformation and blur. We present a deep learning-based solution to the problem of restoring a single turbulence-degraded face image where the amount of geometric distortion and blur at each pixel location is first estimated in terms of variance maps using two separate networks. The estimated variance maps are then used by the Turbulence Distortion Removal Network (TDRN) to restore the image. Furthermore, a confidence-guided image gradient-based loss is proposed to train TDRN. Comprehensive experiments on synthetic and real face images show that the proposed framework is capable of alleviating blur and geometric distortion caused by atmospheric turbulence, and can significantly improve the visual quality. In addition, an ablation study is performed to demonstrate the improvements obtained by different modules in the proposed method.

Large-View 3D Color Face Reconstruction from Dingle Image via UV Location Map and CGAN

Image-based 3D reconstruction has been a hot topic in computer vision research. Since 3D face model can convey more information, it possesses expansive application prospects in comparison with the 2D face image, such as more precise identity information recognition signs, more accurate emotional expression media, etc. In order to reconstruct a realistic 3D color face model from a single large-view 2D face image, a simple but effective method is proposed. First, the encoder-decoder network is applied to generate the 2D UV location map from the original RGB image. Then a simple convolution neural network is applied to regress the 3D face from the UV location map. Finally, a network based on conditional generative adversarial networks (CGAN) is proposed to make up for the missing of UV texture caused by self-occlusion in the case of large pose. Compared with existing 3D face reconstruction methods in Stirling/ESRC 3D Face Database, proposed method can improve the accuracy. More important, a complete and realistic 3D face model can be reconstructed from the large-view face image even in the increasingly complex background.

Prediction of Symptoms on Admission with Early Neurological Deterioration in Single Small Subcortical Infarct.

Early neurological deterioration (END) often occurs during hospitalization in single small subcortical infarct (SSSI). While, symptoms on admission were rarely reported about its performance on predicting the risk of END. The objective of this study is to explore the relationship between symptoms on admission and END in SSSI. Patients with SSSI in the lenticulostriate artery (LSA) territory presenting within 72 hours of stroke onset were screened prospectively. Clinical characteristics, including symptoms on admission, laboratory tests and imaging findings, were collected. Based on the body regions involved including spherical face (SF), upper limb (UL) or lower limb (LL), symptoms on admission were classified into single spherical face (sSF) and any involvement of limbs (AL). END was defined as ≥2 points increase in total National Institutes of Health Stroke Scale (NIHSS) score or ≥ 1 point increase in motor score within 72 hours after admission. Multivariate logistic regression was used to analyze factors associated with END. Out of 5,832 ischemic stroke patients in the database, 394 patients were finally enrolled in analysis. 65 patients (16.5%) developed END. Multivariable logistic regression revealed that symptoms with LL (OR 2.337, 95% CI 1.041-5.246), UL (OR 2.936, 95% CI 1.349-6.390) were both associated with END, while the involvement of SF (OR 0.447, 95% CI 0.249-0.804) showed the opposite result. Further analysis found that symptoms with AL (OR 3.958, 95% CI 1.355-11.565) showed a higher risk of END compared to sSF after adjustment. Our results discovered that symptoms with AL carried a higher risk of END than those involving sSF in SSSI.

Determination of radiated sound power from acoustic sources using the VBSP method and a transparent acoustic boundary

Significant progress has been made to develop a vibration-based sound power (VBSP) method as an alternative means to determine the radiated acoustic energy from structures. This method has been validated for flat plates, cylindrical-shells, simple curved plates, and arbitrarily curved panels. However, many acoustic sources, such as a blender or motor, have sound power contributions that are enclosed within the structure and therefore cannot be scanned properly using a vibrometer. A rigid enclosure with a single mylar face for one side of the enclosure was fabricated to enclose the acoustic source, such as an “acoustic tent” around the structure of interest, so that the VBSP method could be applied to and obtain the radiated sound power. Due to the nature of the sources, this application of the VBSP method will have limitations that are not present in the aforementioned work. A calibration curve was developed to account for the effects that the tent had on the sources used for testing and used to correct the experimental VBSP sound power measurements from the acoustic tent. Results will be shown comparing the sound power obtained from the VBSP method with the sound power of the source in a reverberation chamber using ISO 3741. Funding for this work was provided by the National Science Foundation (NSF). [Funding for this work was provided by the National Science Foundation (NSF).]

Customized Facial Expression Analysis in Video

<i>Existing computer vision-based emotion recognition systems are trained to classify images of faces into a very limited number of emotions. In this work, we take a different approach and train a convolutional neural network (CNN) that is able to distinguish subtle differences in facial expressions appearing in both individual images and videos. For this effect, we train a feature embedding network that maps a facial image into a position in embedding space, such that it is positioned close to similar facial expressions compared with other expressions. We train the feature embedding network using triplet loss on the publicly available facial expression comparison (FEC) data set. The proposed facial expression model is lightweight (4.7M parameters) and obtains a triplet prediction accuracy of 84.5&#x0025;&#x2014;very close to the average human performance of 86.2&#x0025;. A key contribution of this work is adapting the model that has been trained on images to work reliably on video content. To this end, we propose a novel automatic face quality assessment (FaceQA), which allows us to filter out faces that are unsuited for expression analysis. We validate the proposed approach on several applications, including searching video content for a specific expression from a single query face, generating facial expression statistics of various actors appearing in a video, as well as generating facial expression summaries</i>.

Feature Sharing Attention 3D Face Reconstruction with Unsupervised Learning from In-the-Wild Photo Collection

Abstract 3D face reconstruction is an active research topic in the field of computer vision research. It has a wide range of application scenarios. The number of 3D face annotation samples is small and the accuracy of the 3D reconstruction algorithm is insufficient. In this paper, we propose a shared attention module and a feature point constraint mechanism, and proposes a 3D face reconstruction algorithm that takes a single face image as input and uses a encoder-decoder network to predict the 3D face. The prediction view and light network introduces the attention mechanism module, which can improve the feature extraction ability of the network. The albedo and depth introduces shared features to help extract richer facial feature information. At the same time, the self-monitoring method of single image input avoids the high requirements of traditional methods for datasets. We conducted comparative experiments and ablation experiments on the BFM, and CelebA face datasets. The experimental results show that compared with Unsup3d and other representative 3D Face reconstruction algorithms, it has a certain improvement in the two evaluation indexes of Scale Invariant Depth Error (SIDE) and Mean Angle Deviation (MAD).

Pelican Crossing System for Control a Green Man Light with Predicted Age

Traffic lights are generally used to regulate the control flow of traffic at an intersection from all directions, including a pelican crossing system with traffic signals for pedestrians. There are two facilities for walker crossing, namely using a pedestrian bridge and a zebra cross. In general, the traffic signals of the pelican crossing system are a fixed time, whereas other pedestrians need "green man" traffic lights with duration time arrangement. Our research proposes a prototype intelligent pelican crossing system for somebody who crosses the road at zebra crossings, but the risk of falling while crossing is not expected, especially in the elderly age group or pedestrians who are pregnant or carrying children. On the other hand, the problem is that the average step length or stride length (distance in centimeter), cadence or walking rate (in steps per minute), and the possibility of accidents are very high for pedestrians to make sure do crossing during the lights “green man”. The new idea of our research aims to set the adaptive time arrangement on the pelican crossing intelligent system of the traffic lights “green man” based on the age of the pedestrians with artificial intelligence using two combined methods of the FaceNet and AgeNet. The resulting measure can predict the age of pedestrians of the training dataset of 66.67% and testing prototype in real-time with participants on the pelican crossing system of 73% (single face) and 76% (multi faces).

FuseKin: Weighted image fusion based kinship verification under unconstrained age group

Kinship verification using facial images is mainly performed with a single face sample per person. To perform with a single sample, it is very difficult to specify an age group where kin pairs may have higher similarities. To address the above problem, we propose a novel weighted multi sample fusion (WMSF) method. The proposed WMSF method combines kin signals present in multiple samples per person (MSPP) to form a FuseKin image. To select the most discriminant features from the extracted feature vector, we propose a patch based discriminative analysis (PDA) method. Weights are calculated using the PDA method so as to reduce the discrimination between positive FuseKin pairs. Experiments were conducted on two different datasets which contain multiple face image samples per person, namely Family101 and Family in the Wild (FIW) to validate the performance of the proposed methods. Our method achieves competitive results as compared to other state-of-the-art methods.

In-Situ Alloying of CoCrFeNiX High Entropy Alloys by Selective Laser Melting

High Entropy Alloys are a class of alloys which have been shown to largely exhibit stable microstructures, as well as frequently good mechanical properties, particularly when manufactured by additive manufacturing. Due to the large number of potential compositions that their multi-component nature introduces, high throughput alloy development methods are desirable to speed up the investigation of novel alloys. Here, we explore once such method, in-situ alloying during Additive Manufacture, where a powder of a certain pre-alloyed composition is mixed with the required composition of powder of an additional element, such that alloying takes place when powders are melted during the process. To test the effectiveness and capability of the approach, selective laser melting has been used to manufacture pre-alloyed CoCrFeNi, and also CoCrFeNiCu and CoCrFeNiTi alloys by combining pre-alloyed CoCrFeNi powder with elemental powders of Cu and Ti. Processing parameter variations are used to find the highest relative density for each alloy, and samples were then characterised for microstructure and phase composition. The CoCrFeNi alloy shows a single phase face centred cubic (FCC) microstructure, as found with other processing methods. The CoCrFeNiCu alloy has a two phase FCC microstructure with clear partitioning of the Cu, while the CoCrFeNiTi alloy has an FCC matrix phase with NiTi intermetallics and a hexagonal close packed (HCP) phase, as well as unmelted Ti particles. The microstructures therefore differ from those observed in the same alloys manufactured by other methods, mainly due to the presence of areas with higher concentrations than usually encountered of Cu and Ti respectively. Successful in-situ alloying in this process seems to be improved by the added elemental powder having a lower melting point than the base alloy, as well as a low inherent tendency to segregate. While not producing directly comparable microstructures however, the approach does seem to offer advantages for the rapid screening of alloys for AM processability, identifying, for example, extensive solid-state cracking in the CoCrFeNiTi alloy.

Single-Sample Face Recognition Based on Shared Generative Adversarial Network

Single-sample face recognition is a very challenging problem, where each person has only one labeled training sample. It is difficult to describe unknown facial variations. In this paper, we propose a shared generative adversarial network (SharedGAN) to expand the gallery dataset. Benefiting from the shared decoding network, SharedGAN requires only a small number of training samples. After obtaining the generated samples, we join them into a large public dataset. Then, a deep convolutional neural network is trained on the new dataset. We use the well-trained model for feature extraction. With the deep convolutional features, a simple softmax classifier is trained. Our method has been evaluated on AR, CMU-PIE, and FERET datasets. Experimental results demonstrate the effectiveness of SharedGAN and show its robustness for single sample face recognition.