UAVs are commonly used in forest fire detection, but the captured fire images often suffer from blurring due to the rapid motion between the airborne camera and the fire target. In this study, a multi-input, multi-output U-Net architecture that combines spatial domain and frequency domain information is proposed for image deblurring. The architecture includes a multi-branch dilated convolution attention residual module in the encoder to enhance receptive fields and address local features and texture detail limitations. A feature-fusion module integrating spatial frequency domains is also included in the skip connection structure to reduce feature loss and enhance deblurring performance. Additionally, a multi-channel convolution attention residual module in the decoders improves the reconstruction of local and contextual information. A weighted loss function is utilized to enhance network stability and generalization. Experimental results demonstrate that the proposed model outperforms popular models in terms of subjective perception and quantitative evaluation, achieving a PSNR of 32.26 dB, SSIM of 0.955, LGF of 10.93, and SMD of 34.31 on the self-built forest fire datasets and reaching 86% of the optimal PSNR and 87% of the optimal SSIM. In experiments without reference images, the model performs well in terms of LGF and SMD. The results obtained by this model are superior to the currently popular SRN and MPRNet models.
BackgroundOn 20th Jan 2020, a new coronavirus epidemic with "human-to-human" transmission was officially announced, which inevitably caused public panic in China. Perinatal depression (PND) is one of the most common mental health problems. The purpose of this study is to explore the mental impact of COVID-19 outbreak on pregnant women.MethodsA total of 4124 pregnant women during third trimester from 25 hospitals in 10 provinces were recruited in this cross-sectional study, from Jan/01 to Feb/09, 2020. Among them, 1285 pregnant women were assessed after Jan/20, and 2839 were assessed before that time point. The Edinburgh Postnatal Depression Scale (EPDS) was used to evaluate the maternal psychological situation. Prevalence of PND and anxiety were compared between two groups.FindingsPregnant women had higher scores in EPDS (7.7 vs 7.4, P=0.03) and anxiety subscale (3.4 vs 3.2, P=0.04), especially the highest score for thoughts of self-harm (P=0.005) after the declaration of COVID-19 epidemic. Awareness of COVID-19 significantly increased the prevalence of PND (26.0% vs 29.6%, P=0.02). The COVID-19 number of newly- confirmed, suspected infections, and death cases per day was positively associated with the prevalence of PND (P=0.003, 0.004, and 0.001, respectively). Pregnant women those who were underweight, full-time employed, middle income, age < 35 yrs, primiparous, less exercise, appropriate living area appeared to be more susceptible to the outbreak. InterpretationCOVID-19 outbreak can increase the depressive risk of pregnant women, especially in the self-harm inclination, suggesting that psychological intervention is an urgent need for maternal population.Funding: The study was supported by the National Natural Science Foundation of China (81661128010, 81671412), the National Key Research and Development Program of China (2017YFC1001300, 2018YFC1002804, 2016YFC1000203), Foundation of Shanghai Municipal Commission of Health and Family Planning (20144Y0110), Chinese Academy of Medical Sciences Research Unit, Shanghai Jiao Tong University(2019RU056), CAMS Innovation Fund for Medical Sciences (2019-I2M-5-064) and Shanghai Municipal Key Clinical Specialty, Shanghai, China. Conflict of interest: The authors have declared that no conflict of interest exists.Ethical Approval: This study was registered in Chinese Clinical Trial Registry (ChiCTR1900027020) and the ethical approval was obtained from the Institutional Review Board of International Peace Maternity and Child Health Hospital (GKLW2019-11). Informed consent was obtained from all participants before inclusion.
Algal blooms worldwide pose many challenges to drinking water production and pre-oxidation with NaClO, KMnO4 or ozone is commonly used to enhance algal removal. However, these currently employed oxidation processes often result in significant algal cell lysis or impede the operation of subsequent units. Higher algal removal with pre-chlorination process in real water compared to that in synthetic water was accidentally observed. The preliminary results indicated ammonium in real water altered the chlorine to NH2Cl which was responsible for the improved treatment efficiency, which resulted in this study. 1.5~3.0 mg/L of NH2Cl with oxidation time of 3~7 h significantly enhanced algal removal by coagulation. Selective oxidation of surface-adsorbed organic matters (S-AOM) by NH2Cl and peeling them off from the algal surface increased the zeta potential from -20.2 mV to -3.8 mV while promisingly maintaining cell integrity is the primary mechanism of enhancing algal removal by coagulation. The peeled S-AOM kept large molecular weight and acted as polymer aids. Compared to NaClO and KMnO4, NH2Cl showed the best performance in improving algal removal, avoiding cell lysis and decreasing DBPs formation. These venerable virtues suggested that pre-oxidation with NH2Cl is a preferable process for algal-laden water treatment, especially in long-distance water delivery.
Lugol chromoendoscopy is the standard technique to detect an esophageal squamous cell carcinoma (ESCC). However, a high concentration of Lugol's solution can induce mucosal injury and adverse events. We aimed to investigate the optimal concentration of Lugol's solution to reduce mucosal injury and adverse events without degrading image quality.This was a two-phase double-blind randomized controlled trial. In phase I, 200 eligible patients underwent esophagogastroduodenoscopy and then were randomly (1:1:1:1:1) sprayed with 1.2%, 1.0%, 0.8%, 0.6%, or 0.4% Lugol's solution. Image quality, gastric mucosal injury, adverse events, and operation satisfaction were compared to investigate the minimal effective concentration. In phase II, 42 cases of endoscopic mucosectomy for early ESCC were included. The patients were randomly assigned (1:1) to the minimal effective (0.6%) or conventional (1.2%) concentration of Lugol's solution for further comparison of the effectiveness.In phase I, the gastric mucosal injury was significantly reduced in 0.6% group (P < 0.05). Furthermore, there was no statistical significance in image quality between 0.6% and higher concentrations of Lugol's solution (P > 0.05, respectively). It also showed that the operation satisfaction decreased in 1.2% group compared with the lower concentration groups (P < 0.05). In phase II, the complete resection rate was 100% in both groups, while 0.6% Lugol's solution showed higher operation satisfaction (W = 554.500, P = 0.005).The study indicates that 0.6% might be the optimal concentration of Lugol's solution for early detection and delineation of ESCC, considering minimal mucosal injury and satisfied image. The registry of clinical trials: ClinicalTrials.gov (NCT03180944).
The advanced oxidation processes (AOPs) with Fe(II) as homogenous activator suffer from the accumulation of insoluble Fe(III) oxides which results in the decrease of available catalyst and increase of iron sludge. Herein, with Fe(II)/peroxydisulfate (Fe(II)/PDS) process as an example, HSO3– was proposed to enhance the regeneration of Fe(II). The results showed that 250 µM of HSO3– increased the removal of benzoic acid (BA) from 18.8% to 41.3% while further increasing HSO3– dosage promoted its scavenging effects and decreased BA removal. Accordingly, multiple-dosing mode of HSO3– was used which enhanced BA removal to 63.8%. Addition of HSO3– recovered Fe(II) from Fe(III)/Fe(IV) and shifted the distribution of reactive species from Fe(IV) to SO4•−. Besides as the reductant, the reaction of HSO3– with Fe(III) and Fe(IV) severed as the AOPs and produced SO4•−. A mathematic model was constructed for the Fe(II)/PDS/HSO3– process, and it revealed that ~78% of dosed HSO3– was transformed to SO4•–, but ~81% of the generated SO4•− was consumed by HSO3– . The reaction of Fe(IV) with HSO3– and the activation of in situ formed peroxymonosulfate by Fe(II) are chiefly responsible for SO4•− formation with the contributions of ~47% and ~27%, respectively, in the Fe(II)/PDS/HSO3– process. Considering the complexity of the multiple-dosing mode of HSO3– in the real practice, the sparingly soluble CaS2O5 which released HSO3– gradually was synthesized and used. The single dosing of CaS2O5 stood out in terms of promoting pollutant abatement by the Fe(II)/PDS process compared to that adding equivalent of HSO3– . These results will benefit the optimization of the Fe(II)-based AOPs.
The de facto review-involved recommender systems, using review information to enhance recommendation, have received increasing interest over the past years. Thereinto, one advanced branch is to extract salient aspects from textual reviews (i.e., the item attributes that users express) and combine them with the matrix factorization (MF) technique. However, the existing approaches all ignore the fact that semantically different reviews often include opposite aspect information. In particular, positive reviews usually express aspects that users prefer, while the negative ones describe aspects that users dislike. As a result, it may mislead the recommender systems into making incorrect decisions pertaining to user preference modeling. Toward this end, in this article, we present a review polarity-wise recommender model, dubbed as RPR, to discriminately treat reviews with different polarities. To be specific, in this model, positive and negative reviews are separately gathered and used to model the user-preferred and user-rejected aspects, respectively. Besides, to overcome the imbalance of semantically different reviews, we further develop an aspect-aware importance weighting strategy to align the aspect importance for these two kinds of reviews. Extensive experiments conducted on eight benchmark datasets have demonstrated the superiority of our model when compared with several state-of-the-art review-involved baselines. Moreover, our method can provide certain explanations to real-world rating prediction scenarios.
With the increasing investment in the construction of power facilities and the use of higher voltage level, the noise generated by high voltage substation has become an unavoidable issue. The noise of the high-voltage substation is a mixture which is mixed by different noise sources. The mixed noise must be separated to several independent noise signals firstly for researching. An algorithm combined with the Wavelet Transform and the Sparse Component Analysis was presented in this paper which can separate the single channel mixed noise signal of high-voltage substation. The single channel blind source separation can be changed to the underdetermined blind source separation by setting two augmented signals which are the low-frequency part and the high-frequency part separated from the original noise signal by the Wavelet Transform. Then the signals of different noise sources can be separated by the Sparse Component Analysis. It is proved by the simulation and experiment in 330kV high-voltage substation that the algorithm can solve not only the underdetermined blind source separation, but also the single-channel blind source separation effectively.
An efficient solution to the large-scale recommender system is to represent users and items as binary hash codes in the Hamming space. Towards this end, existing methods tend to code users by modeling their Hamming similarities with the items they historically interact with, which are termed as the first-order similarities in this work. Despite their efficiency, these methods suffer from the suboptimal representative capacity, since they forgo the correlation established by connecting multiple first-order similarities, i.e., the relation among the indirect instances, which could be defined as the high-order similarity. To tackle this drawback, we propose to model both the first- and the high-order similarities in the Hamming space through the user-item bipartite graph. Therefore, we develop a novel learning to hash framework, namely Hamming Spatial Graph Convolutional Networks (HS-GCN), which explicitly models the Hamming similarity and embeds it into the codes of users and items. Extensive experiments on three public benchmark datasets demonstrate that our proposed model significantly outperforms several state-of-the-art hashing models, and obtains performance comparable with the real-valued recommendation models.
Organic anion-transporting polypeptides (OATPs) 1B1 and 1B3 are two highly homologous transport proteins. However, OATP1B1- and 1B3-mediated estradiol-17β-glucuronide (E17βG) uptake can be differentially affected by clotrimazole. In this study, by functional characterization on chimeric transporters and single mutants, we find that G45 in transmembrane domain 1 (TM1) and V386 in TM8 are critical for the activation of OATP1B3-mediated E17βG uptake by clotrimazole. However, the effect of clotrimazole on the function of OATP1B3 is substrate-dependent as clotrimazole does not stimulate OATP1B3-mediated uptake of 4′,5′-dibromofluorescein (DBF) and rosuvastatin. In addition, clotrimazole is not transported by OATP1B3, but it can efficiently permeate the plasma membrane due to its lipophilic properties. Homology modeling and molecular docking indicate that E17βG binds in a substrate binding pocket of OATP1B3 through hydrogen bonding and hydrophobic interactions, among which its sterol scaffold forms hydrophobic contacts with V386. In addition, a flexible glycine residue at position 45 is essential for the activation of OATP1B3. Finally, clotrimazole is predicted to bind at an allosteric site, which mainly consists of hydrophobic residues located at the cytoplasmic halves of TMs 4, 5, 10, and 11.
Achieving highly detailed terrain models spanning vast areas is crucial to modern computer graphics. The pipeline for obtaining such terrains is via amplification of a low-resolution terrain to refine the details given a desired theme, which is a time-consuming and labor-intensive process. Recently, data-driven methods, such as the sparse construction tree, have provided a promising direction to equip the artist with better control over the theme. These methods learn to amplify terrain details by using an exemplar of high-resolution detailed terrains to transfer the theme. In this paper, we propose Generative Adversarial Terrain Amplification (GATA) that achieves better local/global coherence compared to the existing data-driven methods while providing even more ways to control the theme. GATA is comprised of two key ingredients. Thefi rst one is a novel embedding of themes into vectors of real numbers to achieve a single tool for multi-theme amplification. The theme component can leverage existing LIDAR data to generate similar terrain features. It can also generate newfi ctional themes by tuning the embedding vector or even encoding a new example terrain into an embedding. The second one is an adversarially trained model that, conditioned on an embedding and a low-resolution terrain, generates a high-resolution terrain adhering to the desired theme. The proposed integral approach reduces the need for unnecessary manual adjustments, can speed up the development, and brings the model quality to a new level. Our implementation of the proposed method has proved successful in large-scale terrain authoring for an open-world game.
'/%3e%3cuse xlink:href='%23a' transform='rotate(23.036 .093 25.536)'/%3e%3cuse xlink:href='%23a' transform='rotate(45.87 1.273 16.18)'/%3e%3cuse xlink:href='%23a' transform='rotate(69.945 .996 12.078)'/%3e%3cuse xlink:href='%23a' transform='rotate(20.66 -19.689 31.932)'/%3e%3c/svg%3e)