Search

Online dynamic hand gesture recognition is challenging mainly due to three reasons: (i) There is no indication when a gesture starts and ends in the video, (ii) performed gestures should only be recognized once, and (iii) the entire architecture should be designed considering the memory and power budget. In this paper, a two-level hierarchical structure consisting of a detector and a classifier is proposed which enables offline-working convolutional neural network (CNN) architectures to operate online efficiently by using sliding window approach. For efficiency analysis, different CNN architectures are applied to compare these architectures over offline classification accuracy, number of parameters and computation complexity. In order to evaluate the single-time activations of the detected gestures, we used Levenshtein distance as an evaluation metric since it can measure misclassifications, multiple detections, and missing detections at the same time. The performance of the approach is evaluated on two public datasets - EgoGesture and NVIDIA Dynamic Hand Gesture Datasets - which require temporal detection and classification of the performed hand gestures. ResNeXt-101 model achieves the state-of-the-art offline classification accuracy of 94.03% on EgoGesture benchmark and competitive results on NVIDIA benchmarks. In online recognition, we obtain very good performances with considerable early detections.

© 2022, The Author(s), under exclusive licence to Springer-Verlag GmbH Germany.Objective: Vaccination is the most efficient way to control the coronavirus disease 2019 (COVID-19) pandemic, but vaccination rates remain below the target level in most countries. This multicenter study aimed to evaluate the vaccination status of hospitalized patients and compare two different booster vaccine protocols. Setting: Inoculation in Turkey began in mid-January 2021. Sinovac was the only available vaccine until April 2021, when BioNTech was added. At the beginning of July 2021, the government offered a third booster dose to healthcare workers and people aged > 50 years who had received the two doses of Sinovac. Of the participants who received a booster, most chose BioNTech as the third dose. Methods: We collected data from 25 hospitals in 16 cities. Patients hospitalized between August 1 and 10, 2021, were included and categorized into eight groups according to their vaccination status. Results: We identified 1401 patients, of which 529 (37.7%) were admitted to intensive care units. Nearly half (47.8%) of the patients were not vaccinated, and those with two doses of Sinovac formed the second largest group (32.9%). Hospitalizations were lower in the group which received 2 doses of Sinovac and a booster dose of BioNTech than in the group which received 3 doses of Sinovac. Conclusion: Effective vaccinations decreased COVID-19-related hospitalizations. The efficacy after two doses of Sinovac may decrease over time; however, it may be enhanced by adding a booster dose. Moreover, unvaccinated patients may be persuaded to undergo vaccination.

Background A considerable percentage of empirical antibiotic treatment fails in hospitalized patients with community-acquired pneumonia (CAP). β-lactams and macrolid (BLM) combination or respiratory fluoroquinolones (FQ) are the most frequently used in these patients. The aim of the present study is to compare the treatment failure (TF) rates in BLM and FQ treatment and to analyze the predictive factors of TF. Method Hospitalized patients who were initially treated with either BLM or FQ were included retrospectively and treatment results of the two regimens were compared. Results Of the 144 patients included in the study, the mean age was 67±16 and 102 (71%) were male. Each group constituted of 72 patients. Antibiotic selection did not alter TF rates, length of stay (LOS) and 30-day mortality. Baseline higher levels of leucocytes, neutrophils to lymphocytes ratio (NLR), C-reactive protein (CRP), BUN/albumin, lactate dehydrogenase/aspartat aminotransferase (LDH/AST) levels and pneumonia severity index (PSI) scores were detected as predictors of TF. Conclusion Empirical treatments with either BLM or FQ do not correlate with TF, LOS and 30-day mortality. NLR, BUN/albumin and LDH/AST may suggest TF. These inexpensive and easily-reachable parameters have the potential as predictors of the treatment outcome in CAP. Notices of retraction Akyil, F. T., Bekir, S. A., Gungor, A., Akyuz, K., Kose, N., Turker, H., Akyil, M., & Sevim, T. (2019). TREATMENT FAILURE AND EMPIRIC ANTIBIOTIC CHOICE FOR HOSPITALIZED PATIENTS WITH COMMUNITY-ACQUIRED PNEUMONIA: Β-LACTAM/Β-LACTAMASE INHIBITOR COMBINED WITH MACROLID OR FLUOROQUINOLONE ALONE?. International Journal of Research -GRANTHAALAYAH, 7(7), 360-369. https://doi.org/10.29121/granthaalayah.v7.i7.2019.778 Article retracted by : EditorReason(s) for retraction : Author's conflicts of interest

As coronavirus disease 2019 (COVID-19) spreads across the world, the ongoing clinical trials are leading to a big race worldwide to develop a treatment that will help control the pandemic. Unfortunately, COVID-19 does not have any known effective treatment with reliable study results yet. In this pandemic, there is not a lot of time to develop a new specific agent because of the rapid spread of the disease. The process of developing a vaccine is long and requires hard work. Although the pathophysiology of the disease is not fully understood, some of the proposed treatment alternatives are based on old evidence and some have been used with the idea that they might work owing to their mechanism of action. The efficacy, reliability, and safety of the currently available treatment alternatives are therefore a matter of debate. Currently, the main therapies used in the treatment of COVID-19 are antiviral drugs and chloroquine/hydroxychloroquine. Other proposed options include tocilizumab, convalescent plasma, and steroids, but the mainstay of the treatment in intensive care units remains supportive therapies. WOS:000600070300012 2-s2.0-85103141655

Scientists from all over the world have been intensively working to discover different aspects of Coronavirus disease 2019 (COVID-19) since the first cluster of cases was reported in China. Herein, we aimed to investigate unclear issues related to transmission and pathogenesis of disease as well as accuracy of diagnostic tests and treatment modalities. A literature search on PubMed, Ovid, and EMBASE databases was conducted, and articles pertinent to identified search terms were extracted. A snow-ball search strategy was followed in order to retrieve additional relevant articles. It was reported that viral spread may occur during the asymptomatic phase of infection, and viral load was suggested to be a useful marker to assess disease severity. In contrast to immune response against viral infections, cytotoxic T lymphocytes decline in SARS-CoV-2 infection, which can be partially explained by direct invasion of T lymphocytes or apoptosis activated by SARS-CoV-2. Dysregulation of the urokinase pathway, cleavage of the SARS-CoV-2 Spike protein by FXa and FIIa, and consumption coagulopathy were the proposed mechanisms of the coagulation dysfunction in COVID-19. False-negative rates of reverse transcriptase polymerase chain reaction varied between 3% and 41% across studies. The probability of the positive test was proposed to decrease with the number of days past from symptom onset. Safety issues related to infection spread limit the use of high flow nasal oxygen (HFNO) and continuous positive airway pressure (CPAP) in hypoxic patients. Further studies are required to elucidate the challenging issues, thus enhancing the management of COVID-19 patients. WOS:000542299000018 2-s2.0-85087118043

Objective: to evaluate long-term effects of COVID-19, and to determine the risk factors in long-COVID in a cohort of the Turkish Thoracic Society (TTS)-TURCOVID multicenter registry.Methods: thirteen centers participated with 831 patients; 504 patients were enrolled after exclusions. The study was designed in three-steps: (1) Phone questionnaire; (2) retrospective evaluation of the medical records; (3) face-to-face visit. Results: in the first step, 93.5% of the patients were hospitalized; 61.7% had a history of pneumonia at the time of diagnosis. A total of 27.1% reported clinical symptoms at the end of the first year. Dyspnea (17.00%), fatigue (6.30%), and weakness (5.00%) were the most prevalent long-term symptoms. The incidence of long-term symptoms was increased by 2.91 fold (95% CI 1.04-8.13, P=0.041) in the presence of chronic obstructive pulmonary disease and by 1.84 fold (95% CI 1.10-3.10, P=0.021) in the presence of pneumonia at initial diagnosis, 3.92 fold (95% Cl 2.29-6.72, P=0.001) of dyspnea and 1.69 fold (95% Cl 1.02-2.80, P=0.040) fatigue persists in the early-post-treatment period and 2.88 fold (95% Cl 1.52- 5.46, P=0.001) in the presence of emergency service admission in the post COVID period. In step 2, retrospective analysis of 231 patients revealed that 1.4% of the chest X-rays had not significantly improved at the end of the first year, while computed tomography (CT) scan detected fibrosis in 3.4%. In step 3, 138 (27.4%) patients admitted to face-to-face visit at the end of first year; at least one symptom persisted in 49.27% patients. The most common symptoms were dyspnea (27.60%), psychiatric symptoms (18.10%), and fatigue (17.40%). Thorax CT revealed fibrosis in 2.4% patients. Conclusions: COVID-19 symptoms can last for extended lengths of time, and severity of the disease as well as the presence of comorbidities might contribute to increased risk. Long-term clinical issues should be regularly evaluated after COVID-19., NA

Many road accidents occur due to distracted drivers. Today, driver monitoring is essential even for the latest autonomous vehicles to alert distracted drivers in order to take over control of the vehicle in case of emergency. In this paper, a spatio-temporal approach is applied to classify drivers' distraction level and movement decisions using convolutional neural networks (CNNs). We approach this problem as action recognition to benefit from temporal information in addition to spatial information. Our approach relies on features extracted from sparsely selected frames of an action using a pre-trained BN-Inception network. Experiments show that our approach outperforms the state-of-the art results on the Distracted Driver Dataset (96.31%), with an accuracy of 99.10% for 10-class classification while providing real-time performance. We also analyzed the impact of fusion using RGB and optical flow modalities with a very recent data level fusion strategy. The results on the Distracted Driver and Brain4Cars datasets show that fusion of these modalities further increases the accuracy., Comment: Accepted for publication by the IEEE Intelligent Transportation Systems Conference (ITSC 2019)

Modern vehicles are equipped with multiple sensors that help engineers to automate driving functions as much as possible. Among other types of sensors, these vehicles are equipped with multiple cameras that enable capturing the video data of the surrounding vehicles. These video data can be later used in training deep neural networks that are typically used within the vehicles to detect objects of interest around them. However, before we can use the video data for training, we must annotate them. Nowadays, the annotation process is typically semi-automated, where the initial annotations are added in a datacenter and then fine-tuned by human annotators. In this paper, we investigate the possibility to utilize high performance computing inside the vehicles in order to add initial annotations at the edge, instead of using the datacenter. With this approach, we expect a more scalable overall solution for the annotation process, because we utilize the computing power that is already available in the vehicles. For this investigation, we focus on heterogeneous automotive platforms of Intel® and analyze their potential for the annotation tasks at the automotive edge.

Acquiring spatio-temporal states of an action is the most crucial step for action classification. In this paper, we propose a data level fusion strategy, Motion Fused Frames (MFFs), designed to fuse motion information into static images as better representatives of spatio-temporal states of an action. MFFs can be used as input to any deep learning architecture with very little modification on the network. We evaluate MFFs on hand gesture recognition tasks using three video datasets - Jester, ChaLearn LAP IsoGD and NVIDIA Dynamic Hand Gesture Datasets - which require capturing long-term temporal relations of hand movements. Our approach obtains very competitive performance on Jester and ChaLearn benchmarks with the classification accuracies of 96.28% and 57.4%, respectively, while achieving state-of-the-art performance with 84.7% accuracy on NVIDIA benchmark., Comment: Accepted to CVPR 2018 as workshop paper

The recent success of emerging RGB-D cameras such as the Kinect sensor depicts a broad prospect of 3-D data-based computer applications. However, due to the lack of a standard testing database, it is difficult to evaluate how the face recognition technology can benefit from this up-to-date imaging sensor. In order to establish the connection between the Kinect and face recognition research, in this paper, we present the first publicly available face database (i.e., KinectFaceDB 1 ) based on the Kinect sensor. The database consists of different data modalities (well-aligned and processed 2-D, 2.5-D, 3-D, and video-based face data) and multiple facial variations. We conducted benchmark evaluations on the proposed database using standard face recognition methods, and demonstrated the gain in performance when integrating the depth data with the RGB data via score-level fusion. We also compared the 3-D images of Kinect (from the KinectFaceDB) with the traditional high-quality 3-D scans (from the FRGC database) in the context of face biometrics, which reveals the imperative needs of the proposed database for face recognition research. 1 Online at http://rgb-d.eurecom.fr

In this paper a spectral decomposition-based method for interharmonic computation is proposed for power systems where the fundamental frequency fluctuates significantly. In the proposed method, the frequency domain components of the voltage waveform are obtained by Kalman filtering. Both the system fundamental frequency and the correct spectrum of the voltage waveform, and hence the exact interharmonics are obtained. The proposed method is tested with both simulated and field data obtained from different electric arc furnace (EAF) plants, where the system frequency deviates continuously due to the fluctuating load demands. Since the interharmonic frequencies are obtained by using Kalman filtering, no leakage effect of the DFT-based methods is involved in case of frequency deviations, which is an important advantage of the proposed method.

International audience; Recent studies show that the performances offace recognition systems degrade in presence of makeup onface. In this paper, a facial makeup detector is proposed tofurther reduce the impact of makeup in face recognition.The performance of the proposed technique is tested usingthree publicly available facial makeup databases. The proposedtechnique extracts a feature vector that captures the shapeand texture characteristics of the input face. After featureextraction, two types of classifiers (i.e. SVM and Alligator) areapplied for comparison purposes. In this study, we observed thatboth classifiers provide significant makeup detection accuracy.There are only few studies regarding facial makeup detectionin the state-of-the art. The proposed technique is novel andoutperforms the state-of-the art significantly.

Automatic face recognition, required in law enforcement applications such as surveillance, border security and forensic investigation, is a process in which an individual is identified or verified based on facial characteristics. Researchers have proposed several algorithms that can effectively recognize individuals in controlled environment with minor variations in pose, expression, and illumination (Zhao et al., 2003), (Li & Jain, 2005), (Wechsler, 2006), (Delac & Grgic, 2007). In recent face recognition test reports such as FRVT 2002, FRGC 2004, and FRGC 2006 (Philips et al., 2006 & Philips et al., 2007), the results show that under normal changes in constrained environment, the performance of existing face recognition systems is greatly enhanced. However, in most real world applications, images may not be of good quality or user may not be cooperative or there may be temporal variations and dissimilarities in facial characteristics that are artificially created using disguise accessories. Challenges in automatic face recognition can be classified into six categories: illumination, image quality, expression, pose, aging, and disguise. Among these challenges, recognition of faces with disguise is a major challenge and has only been recently addressed by few researchers (Alexander & Smith, 2003), (Ramanathan et al, 2004), (Silva & Rosa, 2003), (Singh et al., 2008). As shown in Fig. 1, the inter-personal and intra-personal characteristics can be modeled using disguise accessories to alter the appearance of an individual, to impersonate another person, or to hide one’s identity. For example, a criminal can alter facial features and appearance using makeup tools and accessories to remain elusive from law enforcement. The challenges due to disguise cause change in visual perception, alter actual data, make pertinent facial information disappear, mask features to varying degrees, or introduce extraneous artifacts in the face image. Existing face recognition algorithms may not be able to provide the desired level of security for such cases. In literature, Ramanathan et al. (Ramanathan et al., 2004) studied facial similarity for several variations including disguise by forming two eigenspaces from two halves of the face, one using the left half and other using the right half. From the test image, optimally illuminated half face is chosen and is projected into the eigenspace. This algorithm has been tested on the AR face database (Martinez & Benavente, 1998) and the National Geographic database (Ramanathan et al., 2004) which consists of variations in smile, glasses, and illumination. An accuracy of around 39% for best two matches is reported on the AR database. Silva and Rosa O pe n A cc es s D at ab as e w w w .in te ch w eb .o rg

Facial cosmetics, also called makeup, may change the appearance of a face that we could perceive. In order to contribute to studies in image processing related to facial cosmetics, a database is built which contains multiple images per person with and without applied cosmetics. Annotations provide detailed information about the amount and location of applied makeup for each picture. Furthermore, a classification approach is presented. The classification approach takes the altering effect of the applied cosmetics and the application area into account. Since automatic face recognition evolved to an important topic over the last decades and is affected by facial cosmetics, preliminary tests are done to evaluate their impact on automatic face recognition. The face as a whole as well as its most significant makeup application areas that are skin, eyes and mouth are investigated separately.

In this paper, a new face anti-spoofing approach, which is based on analysis of contrast and texture characteristics of captured and recaptured images, is proposed to detect photograph spoofing. Since photo image is a recaptured image, it may show quite different contrast and texture characteristics when compared to a real face image. In a spoofing attempt, image rotation is quite possible. Therefore, in this paper, a rotation invariant local binary pattern variance (LBPV) based method is selected to be used. The approach is tested on the publicly available NUAA photo-impostor database, which is constructed under illumination and place change. The results show that the approach is competitive with other existing methods tested on the same database. It is especially useful for conditions when photos are held by hand to spoof the system. Since an LBPV based method is used, it is robust to illumination changes. It is non-intrusive and simple.

Numerous major challenges in face recognition, such as pose, illumination, expression and aging, have been investigated extensively. All those variations modify the texture and/or the shape of the face in a similar manner for different individuals. However, studies on alterations applied on face via plastic surgery or prosthetic make-up which can be in countless different ways and amounts, are still very limited. In this paper, we analyze how such changes on nose region affect the face recognition performances of several key techniques. For this purpose, a simulated face database is prepared using FRGC v1.0 in which nose in each sample is replaced with another randomly chosen one. Since this is a 3D database, the impact analysis is not limited to only 2D, which is one of the novelties of this study. Performance comparisons of three 2D and four 3D algorithms are provided. In addition, differently from previous works, baseline results for the original database are also reported. Hence, the impact which is purely due to the applied nose alterations can be measured. The experimental results indicate that with the introduction of alterations both modalities lose precision, especially 3D.