Face Detection and Recognition Method Based on Improved Convolutional Neural Network

with rapid development of deep learning technology, face recognition based on deep convolutional neural network becomes one of the main research methods. In order to solve the problems of information loss and equal treatment of each element in the input feature graph in the traditional pooling method of convolutional neural network, a face recognition algorithm based on convolutional neural network is proposed in this paper. First, MTCNN algorithm is used to detect the faces and do gray processing, and then a local weighted average pooling method based on local concern strategy is designed and a convolutional neural network based on VGG16 to recognize faces is constructed which is finally compared with common convolutional neural network. The experimental results show that this method has good face recognition accuracy in common face databases.

Microstructure analysis and image-based modelling of face masks for COVID-19 virus protection

SARS-CoV-2 may spread through respiratory droplets released by infected individuals. The viruses are transmitted in moist droplets which cause coronavirus disease. Many countries have mandated the wearing of face masks, to various extents. However, the efficacy of masks is yet to be well rationalised given the limited microstructure information. Here, three common face masks and associated air permeations were revealed by coupling X-ray tomographic imaging and infrared thermal imaging techniques. Quantitative parameters have been extracted from the 3D images. Also, image-based modelling was performed to simulate the permeability to show how droplets pass through the porous structure. Our results show that the N95 mask has the smallest average pore diameter (~30 µm) and the densest nanoscale fibres which provides superior droplet filtration among all cases. Modifications to the N95 masks are proposed to develop the next generation mask with higher efficacy and better breathability.

Real-time mask detection and face recognition using eigenfaces and local binary pattern histogram for attendance system

Face recognition is gaining popularity as one of the biometrics methods for an attendance system in an organization. Due to the pandemic, the common face recognition system needs to be modified to meet the current needs, whereby facemask detection is necessary. The main objective of this paper is to investigate and develop a real-time face recognition system for the attendance system based on the current scenarios. The proposed framework consists of face detection, mask detection, face recognition, and attendance report generation modules. The face and facemask detection is performed using the haar cascade classifier. Two techniques for face recognition were investigated, the eigenfaces and local binary pattern histogram. The initial experimental results and implementation at Kuching Community College show the effectiveness of the system. For future work, an approach that is able to perform masked face recognition will be investigated.

Strategies to minimize SARS-CoV-2 transmission in classroom settings: Combined impacts of ventilation and mask effective filtration efficiency

The impact of the COVID-19 pandemic continues to be significant and global. As the global community learns more about the novel coronavirus SARS-CoV-2, there is strong evidence that a significant modality of transmission is via the long-range airborne route, referred to here as aerosol transmission. In this paper, we evaluate the efficacy of ventilation, mask effective filtration efficiency, and the combined effect of the two on the reduction of aerosol infection probability for COVID-19 in a classroom setting. The Wells-Riley equation is used to predict the conditional probability of infection for three distinct airborne exposure scenarios: (1) an infectious instructor exposing susceptible students; (2) an infectious student exposing other susceptible students; and (3) an infectious student exposing a susceptible instructor. Field measurements were performed in a classroom using a polydisperse neutralized salt (NaCl) aerosol, generated in a size range consistent with human-generated SARS-CoV-2 containing bioaerosols, as a safe surrogate. Measurements included time-resolved and size-resolved NaCl aerosol concentration distributions and size-resolved effective filtration efficiency of different masks with and without mask fitters. The measurements were used to validate assumptions and inputs for the Wells-Riley model. Aerosol dynamics and distribution measurements confirmed that the majority of the classroom space is uniform in aerosol concentration within a factor of 2 or better for distances > 2 m from the aerosol source. Mask effective filtration efficiency measurements show that most masks fit poorly with estimated leakage rates typically > 50%, resulting in significantly reduced effective filtration efficiency. However, effective filtration efficiencies approaching the mask material filtration efficiency were achievable using simple mask fitters. Wells-Riley model results for the different scenarios suggest that ventilation of the classroom alone is not able to achieve infection probabilities less than 0.01 (1%) for air exchanges rates up to 10 h−1 and an event duration of one hour. The use of moderate to high effective filtration efficiency masks by all individuals present, on the other hand, was able to significantly reduce infection probability and could achieve reductions in infection probability by 5x, 10x, or even >100x dependent on the mask used and use of a mask fitter. This enables conditional infection probabilities < 0.001 (0.1%) or even < 0.0001 (0.01%) to be reached with the use of masks and mask fitters alone. Finally, the results demonstrate that the reductions provided by ventilation and masks are synergistic and multiplicative. The results reinforce the use of properly donned masks to achieve reduced aerosol transmission of SARS-CoV-2 and other infectious diseases transmitted via respiratory aerosol indoors and provide new motivation to further improve the effective filtration efficiency of common face coverings through improved design, and/or the use of mask fitters.

Performance Tuning Techniques for Face Detection Algorithms on GPGPU

Face detection algorithms varies in speed and performance on GPUs. Different algorithms can report different speeds on different GPUs that are not governed by linear or nearlinear approximations. This is due to many factors such as register file size, occupancy rate of the GPU, speed of the memory, and speed of double precision processors. This paper studies the most common face detection algorithms LBP and Haar-like and study the bottlenecks associated with deploying both algorithms on different GPU architectures. The study focuses on the bottlenecks and the associated techniques to resolve them based on the different GPUs specifications.

Structural Study of Metal Binding and Coordination in Ancient Metallo-β-Lactamase PNGM-1 Variants

The increasing incidence of community- and hospital-acquired infections with multidrug-resistant (MDR) bacteria poses a critical threat to public health and the healthcare system. Although β-lactam antibiotics are effective against most bacterial infections, some bacteria are resistant to β-lactam antibiotics by producing β-lactamases. Among β-lactamases, metallo-β-lactamases (MBLs) are especially worrisome as only a few inhibitors have been developed against them. In MBLs, the metal ions play an important role as they coordinate a catalytic water molecule that hydrolyzes β-lactam rings. We determined the crystal structures of different variants of PNGM-1, an ancient MBL with additional tRNase Z activity. The variants were generated by site-directed mutagenesis targeting metal-coordinating residues. In PNGM-1, both zinc ions are coordinated by six coordination partners in an octahedral geometry, and the zinc-centered octahedrons share a common face. Structures of the PNGM-1 variants confirm that the substitution of a metal-coordinating residue causes the loss of metal binding and β-lactamase activity. Compared with PNGM-1, subclass B3 MBLs lack one metal-coordinating residue, leading to a shift in the metal-coordination geometry from an octahedral to tetrahedral geometry. Our results imply that a subtle change in the metal-binding site of MBLs can markedly change their metal-coordination geometry and catalytic activity.

DTFA-Net: Dynamic and Texture Features Fusion Attention Network for Face Antispoofing

For face recognition systems, liveness detection can effectively avoid illegal fraud and improve the safety of face recognition systems. Common face attacks include photo printing and video replay attacks. This paper studied the differences between photos, videos, and real faces in static texture and motion information and proposed a living detection structure based on feature fusion and attention mechanism, Dynamic and Texture Fusion Attention Network (DTFA-Net). We proposed a dynamic information fusion structure of an interchannel attention block to fuse the magnitude and direction of optical flow to extract facial motion features. In addition, for the face detection failure of HOG algorithm under complex illumination, we proposed an improved Gamma image preprocessing algorithm, which effectively improved the face detection ability. We conducted experiments on the CASIA-MFSD and Replay Attack Databases. According to experiments, the DTFA-Net proposed in this paper achieved 6.9% EER on CASIA and 2.2% HTER on Replay Attack that was comparable to other methods.

Ventilating the Bearded: A Randomized Crossover Trial Comparing a Novel Bag-Valve-Guedel Adaptor to a Standard Mask

Introduction: Emergency field ventilation using bag-valve face mask devices can be difficult to perform, especially in bearded individuals. In view of the increasing numbers of servicemen and civilians sporting a beard or moustache, the issue of finding a technical solution for ventilation in this population has gained importance. We therefore developed a novel adaptor that enables the direct connection of a bag-valve device to a Guedel-type oropharyngeal airway device thereby directly connecting the oral airway to the bag valve, eliminating the need for a face mask. The objective of this study was to compare the efficacy of the bag-valve-Guedel adaptor (BVGA) to the common face mask in healthy bearded volunteers. Methods: This study was a randomized-by-sequence, crossover-controlled trial (NCT02768246) approved by the local IRB (0051-16-HMO). All subjects signed an informed consent before participation. Twenty-five healthy bearded men (age 28 ± 7) were recruited. After randomization, the first group (mask then BVGA, n = 12) began breathing room air through the face mask, followed by 100% O2. After washout in room air, the procedure was repeated with the BVGA. The second group (BVGA then mask, n = 13) began with the BVGA followed by the face mask. Subjects were awake and breathed spontaneously throughout the experiment. Therefore, a Guedel was not used. Physiological and respiratory parameters were monitored continuously. The primary endpoint was the presence of suspected leak as determined by end-tidal-CO2 (EtCO2 &lt; 20 mmHg). Secondary endpoints included tidal volume and safety. Results: The order of device use did not affect the results significantly (p &gt; 0.05 by Mann-Whitney-U test); therefore, the data were pooled. There were no cases of suspected leak while breathing through the BVGA. By contrast, while breathing through a face mask, there were 8 of 25 (32%) and 5 of 25 (20%) cases of suspected leak in air and 100% O2, respectively (air: p = 0.002; 100% O2: p = 0.014 by McNemar test). No adverse events were observed. Conclusions: In bearded individuals, the BVGA provides significantly more efficient (less leak) ventilation compared to a face mask. This is also of particular importance in view of the increasing number of bearded individuals serving in the armed forces. Moreover, since effective ventilation with a mask requires experience, the relatively easy-to-apply BVGA will enable less experienced first responders to achieve higher success rates in this critical phase of treatment. Further studies are planned to evaluate the efficacy of the BVGA in the prehospital setting.

Systemic cohesion in social media conversations: Cases on Facebook and Twitter

As a new way of communication, social media conversations on the Internet do have some characteristics that are different from common face-to-face conversations. One of the differences lies in the existence of systemic cohesion in addition to the established conventional cohesion. Systemic cohesion is a form of structural and textual unity which is generated by the system of a social media platform and is not available in offline discourse. This article is aimed at describing the phenomena of systemic cohesion of social media conversations, particularly on Facebook and Twitter, by analyzing them based on the classification of cohesive devices made by Halliday and Hasan (1976, 1985). In general, systemic cohesion can be distinguished into structural and textual cohesion. The former is represented by the functions of conversation components, indentation, and vertical line whereas the latter is implemented in the form of mention and hashtag. Facebook and Twitter share both similarities and differences in either case.