Showing papers by "Nankai University published in 2021"

Res2Net: A New Multi-Scale Backbone Architecture

Abstract: Representing features at multiple scales is of great importance for numerous vision tasks. Recent advances in backbone convolutional neural networks (CNNs) continually demonstrate stronger multi-scale representation ability, leading to consistent performance gains on a wide range of applications. However, most existing methods represent the multi-scale features in a layer-wise manner. In this paper, we propose a novel building block for CNNs, namely Res2Net, by constructing hierarchical residual-like connections within one single residual block. The Res2Net represents multi-scale features at a granular level and increases the range of receptive fields for each network layer. The proposed Res2Net block can be plugged into the state-of-the-art backbone CNN models, e.g., ResNet, ResNeXt, and DLA. We evaluate the Res2Net block on all these models and demonstrate consistent performance gains over baseline models on widely-used datasets, e.g., CIFAR-100 and ImageNet. Further ablation studies and experimental results on representative computer vision tasks, i.e., object detection, class activation mapping, and salient object detection, further verify the superiority of the Res2Net over the state-of-the-art baseline methods. The source code and trained models are available on https://mmcheng.net/res2net/ .

Guidelines for the use and interpretation of assays for monitoring autophagy (4th edition)

Abstract: In 2008, we published the first set of guidelines for standardizing research in autophagy. Since then, this topic has received increasing attention, and many scientists have entered the field. Our knowledge base and relevant new technologies have also been expanding. Thus, it is important to formulate on a regular basis updated guidelines for monitoring autophagy in different organisms. Despite numerous reviews, there continues to be confusion regarding acceptable methods to evaluate autophagy, especially in multicellular eukaryotes. Here, we present a set of guidelines for investigators to select and interpret methods to examine autophagy and related processes, and for reviewers to provide realistic and reasonable critiques of reports that are focused on these processes. These guidelines are not meant to be a dogmatic set of rules, because the appropriateness of any assay largely depends on the question being asked and the system being used. Moreover, no individual assay is perfect for every situation, calling for the use of multiple techniques to properly monitor autophagy in each experimental setting. Finally, several core components of the autophagy machinery have been implicated in distinct autophagic processes (canonical and noncanonical autophagy), implying that genetic approaches to block autophagy should rely on targeting two or more autophagy-related genes that ideally participate in distinct steps of the pathway. Along similar lines, because multiple proteins involved in autophagy also regulate other cellular pathways including apoptosis, not all of them can be used as a specific marker for bona fide autophagic responses. Here, we critically discuss current methods of assessing autophagy and the information they can, or cannot, provide. Our ultimate goal is to encourage intellectual and technical innovation in the field.

Pyramid Vision Transformer: A Versatile Backbone for Dense Prediction without Convolutions

Abstract: Although using convolutional neural networks (CNNs) as backbones achieves great successes in computer vision, this work investigates a simple backbone network useful for many dense prediction tasks without convolutions. Unlike the recently-proposed Transformer model (e.g., ViT) that is specially designed for image classification, we propose Pyramid Vision Transformer~(PVT), which overcomes the difficulties of porting Transformer to various dense prediction tasks. PVT has several merits compared to prior arts. (1) Different from ViT that typically has low-resolution outputs and high computational and memory cost, PVT can be not only trained on dense partitions of the image to achieve high output resolution, which is important for dense predictions but also using a progressive shrinking pyramid to reduce computations of large feature maps. (2) PVT inherits the advantages from both CNN and Transformer, making it a unified backbone in various vision tasks without convolutions by simply replacing CNN backbones. (3) We validate PVT by conducting extensive experiments, showing that it boosts the performance of many downstream tasks, e.g., object detection, semantic, and instance segmentation. For example, with a comparable number of parameters, RetinaNet+PVT achieves 40.4 AP on the COCO dataset, surpassing RetinNet+ResNet50 (36.3 AP) by 4.1 absolute AP. We hope PVT could serve as an alternative and useful backbone for pixel-level predictions and facilitate future researches. Code is available at this https URL.

Rethinking RGB-D Salient Object Detection: Models, Data Sets, and Large-Scale Benchmarks

Abstract: The use of RGB-D information for salient object detection (SOD) has been extensively explored in recent years. However, relatively few efforts have been put toward modeling SOD in real-world human activity scenes with RGB-D. In this article, we fill the gap by making the following contributions to RGB-D SOD: 1) we carefully collect a new S al i ent P erson (SIP) data set that consists of ~1 K high-resolution images that cover diverse real-world scenes from various viewpoints, poses, occlusions, illuminations, and background s; 2) we conduct a large-scale (and, so far, the most comprehensive) benchmark comparing contemporary methods, which has long been missing in the field and can serve as a baseline for future research, and we systematically summarize 32 popular models and evaluate 18 parts of 32 models on seven data sets containing a total of about 97k images; and 3) we propose a simple general architecture, called deep depth-depurator network (D3Net). It consists of a depth depurator unit (DDU) and a three-stream feature learning module (FLM), which performs low-quality depth map filtering and cross-modal feature learning, respectively. These components form a nested structure and are elaborately designed to be learned jointly. D3Net exceeds the performance of any prior contenders across all five metrics under consideration, thus serving as a strong model to advance research in this field. We also demonstrate that D3Net can be used to efficiently extract salient object masks from real scenes, enabling effective background-changing application with a speed of 65 frames/s on a single GPU. All the saliency maps, our new SIP data set, the D3Net model, and the evaluation tools are publicly available at https://github.com/DengPingFan/D3NetBenchmark .

Direct Self‐Assembly of MXene on Zn Anodes for Dendrite‐Free Aqueous Zinc‐Ion Batteries

Abstract: Metallic zinc is a promising anode candidate of aqueous zinc-ion batteries owing to its high theoretical capacity and low redox potential. However, Zn anodes usually suffer from dendrite and side reactions, which will degrade their cycle stability and reversibility. Herein, we developed an in situ spontaneously reducing/assembling strategy to assemble a ultrathin and uniform MXene layer on the surface of Zn anodes. The MXene layer endows the Zn anode with a lower Zn nucleation energy barrier and a more uniformly distributed electric field through the favorable charge redistribution effect in comparison with pure Zn. Therefore, MXene-integrated Zn anode exhibits obviously low voltage hysteresis and excellent cycling stability with dendrite-free behaviors, ensuring the high capacity retention and low polarization potential in zinc-ion batteries.

Advances and Challenges for the Electrochemical Reduction of CO2 to CO: From Fundamentals to Industrialization

Abstract: The electrochemical carbon dioxide reduction reaction (CO2 RR) provides an attractive approach to convert renewable electricity into fuels and feedstocks in the form of chemical bonds. Among the different CO2 RR pathways, the conversion of CO2 into CO is considered one of the most promising candidate reactions because of its high technological and economic feasibility. Integrating catalyst and electrolyte design with an understanding of the catalytic mechanism will yield scientific insights and promote this technology towards industrial implementation. Herein, we give an overview of recent advances and challenges for the selective conversion of CO2 into CO. Multidimensional catalyst and electrolyte engineering for the CO2 RR are also summarized. Furthermore, recent studies on the large-scale production of CO are highlighted to facilitate industrialization of the electrochemical reduction of CO2 . To conclude, the remaining technological challenges and future directions for the industrial application of the CO2 RR to generate CO are highlighted.

JCS: An Explainable COVID-19 Diagnosis System by Joint Classification and Segmentation

Abstract: Recently, the coronavirus disease 2019 (COVID-19) has caused a pandemic disease in over 200 countries, influencing billions of humans. To control the infection, identifying and separating the infected people is the most crucial step. The main diagnostic tool is the Reverse Transcription Polymerase Chain Reaction (RT-PCR) test. Still, the sensitivity of the RT-PCR test is not high enough to effectively prevent the pandemic. The chest CT scan test provides a valuable complementary tool to the RT-PCR test, and it can identify the patients in the early-stage with high sensitivity. However, the chest CT scan test is usually time-consuming, requiring about 21.5 minutes per case. This paper develops a novel Joint Classification and Segmentation ( JCS ) system to perform real-time and explainable COVID- 19 chest CT diagnosis. To train our JCS system, we construct a large scale COVID- 19 Classification and Segmentation ( COVID-CS ) dataset, with 144,167 chest CT images of 400 COVID- 19 patients and 350 uninfected cases. 3,855 chest CT images of 200 patients are annotated with fine-grained pixel-level labels of opacifications, which are increased attenuation of the lung parenchyma. We also have annotated lesion counts, opacification areas, and locations and thus benefit various diagnosis aspects. Extensive experiments demonstrate that the proposed JCS diagnosis system is very efficient for COVID-19 classification and segmentation. It obtains an average sensitivity of 95.0% and a specificity of 93.0% on the classification test set, and 78.5% Dice score on the segmentation test set of our COVID-CS dataset. The COVID-CS dataset and code are available at https://github.com/yuhuan-wu/JCS .

Horizontally Arranged Zinc Platelet Electrodeposits Modulated by Fluorinated Covalent Organic Framework Film for High-Rate and Durable Aqueous Zinc Ion Batteries

Abstract: Rechargeable aqueous zinc-ion batteries (RZIBs) provide a promising complementarity to the existing lithium-ion batteries due to their low cost, non-toxicity and intrinsic safety. However, Zn anodes suffer from zinc dendrite growth and electrolyte corrosion, resulting in poor reversibility. Here, we develop an ultrathin, fluorinated two-dimensional porous covalent organic framework (FCOF) film as a protective layer on the Zn surface. The strong interaction between fluorine (F) in FCOF and Zn reduces the surface energy of the Zn (002) crystal plane, enabling the preferred growth of (002) planes during the electrodeposition process. As a result, Zn deposits show horizontally arranged platelet morphology with (002) orientations preferred. Furthermore, F-containing nanochannels facilitate ion transport and prevent electrolyte penetration for improving corrosion resistance. The FCOF@Zn symmetric cells achieve stability for over 750 h at an ultrahigh current density of 40 mA cm−2. The high-areal-capacity full cells demonstrate hundreds of cycles under high Zn utilization conditions. Rechargeable aqueous zinc-ion batteries are promising but the zinc anode suffers from dendrite growth and electrolyte corrosion. Here, the authors develop a fluorinated covalent organic framework film as a protective layer for aqueous zinc anode battery.

Salmonella Typhimurium reprograms macrophage metabolism via T3SS effector SopE2 to promote intracellular replication and virulence

Abstract: Salmonella Typhimurium establishes systemic infection by replicating in host macrophages Here we show that macrophages infected with S Typhimurium exhibit upregulated glycolysis and decreased serine synthesis, leading to accumulation of glycolytic intermediates The effects on serine synthesis are mediated by bacterial protein SopE2, a type III secretion system (T3SS) effector encoded in pathogenicity island SPI-1 The changes in host metabolism promote intracellular replication of S Typhimurium via two mechanisms: decreased glucose levels lead to upregulated bacterial uptake of 2- and 3-phosphoglycerate and phosphoenolpyruvate (carbon sources), while increased pyruvate and lactate levels induce upregulation of another pathogenicity island, SPI-2, known to encode virulence factors Pharmacological or genetic inhibition of host glycolysis, activation of host serine synthesis, or deletion of either the bacterial transport or signal sensor systems for those host glycolytic intermediates impairs S Typhimurium replication or virulence Salmonella Typhimurium establishes systemic infection by replicating in host macrophages Here, Jiang et al show that infected macrophages exhibit upregulated glycolysis and decreased serine synthesis, leading to accumulation of glycolytic intermediates that promote intracellular replication and virulence of S Typhimurium

Revisiting Video Saliency Prediction in the Deep Learning Era

Abstract: Predicting where people look in static scenes, a.k.a visual saliency, has received significant research interest recently. However, relatively less effort has been spent in understanding and modeling visual attention over dynamic scenes. This work makes three contributions to video saliency research. First, we introduce a new benchmark, called DHF1K (Dynamic Human Fixation 1K), for predicting fixations during dynamic scene free-viewing, which is a long-time need in this field. DHF1K consists of 1K high-quality elaborately-selected video sequences annotated by 17 observers using an eye tracker device. The videos span a wide range of scenes, motions, object types and backgrounds. Second, we propose a novel video saliency model, called ACLNet (Attentive CNN-LSTM Network), that augments the CNN-LSTM architecture with a supervised attention mechanism to enable fast end-to-end saliency learning. The attention mechanism explicitly encodes static saliency information, thus allowing LSTM to focus on learning a more flexible temporal saliency representation across successive frames. Such a design fully leverages existing large-scale static fixation datasets, avoids overfitting, and significantly improves training efficiency and testing performance. Third, we perform an extensive evaluation of the state-of-the-art saliency models on three datasets : DHF1K, Hollywood-2, and UCF sports. An attribute-based analysis of previous saliency models and cross-dataset generalization are also presented. Experimental results over more than 1.2K testing videos containing 400K frames demonstrate that ACLNet outperforms other contenders and has a fast processing speed (40 fps using a single GPU). Our code and all the results are available at https://github.com/wenguanwang/DHF1K .

Endothelial Dysfunction in Atherosclerotic Cardiovascular Diseases and Beyond: From Mechanism to Pharmacotherapies.

Abstract: The endothelium, a cellular monolayer lining the blood vessel wall, plays a critical role in maintaining multiorgan health and homeostasis. Endothelial functions in health include dynamic maintenance of vascular tone, angiogenesis, hemostasis, and the provision of an antioxidant, anti-inflammatory, and antithrombotic interface. Dysfunction of the vascular endothelium presents with impaired endothelium-dependent vasodilation, heightened oxidative stress, chronic inflammation, leukocyte adhesion and hyperpermeability, and endothelial cell senescence. Recent studies have implicated altered endothelial cell metabolism and endothelial-to-mesenchymal transition as new features of endothelial dysfunction. Endothelial dysfunction is regarded as a hallmark of many diverse human panvascular diseases, including atherosclerosis, hypertension, and diabetes. Endothelial dysfunction has also been implicated in severe coronavirus disease 2019. Many clinically used pharmacotherapies, ranging from traditional lipid-lowering drugs, antihypertensive drugs, and antidiabetic drugs to proprotein convertase subtilisin/kexin type 9 inhibitors and interleukin 1β monoclonal antibodies, counter endothelial dysfunction as part of their clinical benefits. The regulation of endothelial dysfunction by noncoding RNAs has provided novel insights into these newly described regulators of endothelial dysfunction, thus yielding potential new therapeutic approaches. Altogether, a better understanding of the versatile (dys)functions of endothelial cells will not only deepen our comprehension of human diseases but also accelerate effective therapeutic drug discovery. In this review, we provide a timely overview of the multiple layers of endothelial function, describe the consequences and mechanisms of endothelial dysfunction, and identify pathways to effective targeted therapies. SIGNIFICANCE STATEMENT: The endothelium was initially considered to be a semipermeable biomechanical barrier and gatekeeper of vascular health. In recent decades, a deepened understanding of the biological functions of the endothelium has led to its recognition as a ubiquitous tissue regulating vascular tone, cell behavior, innate immunity, cell-cell interactions, and cell metabolism in the vessel wall. Endothelial dysfunction is the hallmark of cardiovascular, metabolic, and emerging infectious diseases. Pharmacotherapies targeting endothelial dysfunction have potential for treatment of cardiovascular and many other diseases.

Smoothing the energy transfer pathway in quasi-2D perovskite films using methanesulfonate leads to highly efficient light-emitting devices.

Abstract: Quasi-two-dimensional (quasi-2D) Ruddlesden–Popper (RP) perovskites such as BA2Csn–1PbnBr3n+1 (BA = butylammonium, n > 1) are promising emitters, but their electroluminescence performance is limited by a severe non-radiative recombination during the energy transfer process. Here, we make use of methanesulfonate (MeS) that can interact with the spacer BA cations via strong hydrogen bonding interaction to reconstruct the quasi-2D perovskite structure, which increases the energy acceptor-to-donor ratio and enhances the energy transfer in perovskite films, thus improving the light emission efficiency. MeS additives also lower the defect density in RP perovskites, which is due to the elimination of uncoordinated Pb2+ by the electron-rich Lewis base MeS and the weakened adsorbate blocking effect. As a result, green light-emitting diodes fabricated using these quasi-2D RP perovskite films reach current efficiency of 63 cd A−1 and 20.5% external quantum efficiency, which are the best reported performance for devices based on quasi-2D perovskites so far. Owing to large exciton binding energy, quasi-2D perovskite is promising for light-emitting application, yet inhomogeneous phases distribution limits the potential. Here, the authors improve the performance by using MeS additive to regulate the phase distribution and to reduce defect density in the films.

Non-concentrated aqueous electrolytes with organic solvent additives for stable zinc batteries.

Abstract: Rechargeable aqueous zinc batteries (RAZBs) are promising for large-scale energy storage because of their superiority in addressing cost and safety concerns. However, their practical realization is hampered by issues including dendrite growth, poor reversibility and low coulombic efficiency (CE) of Zn anodes due to parasitic reactions. Here, we report a non-concentrated aqueous electrolyte composed of 2 m zinc trifluoromethanesulfonate (Zn(OTf)2) and the organic dimethyl carbonate (DMC) additive to stabilize the Zn electrochemistry. Unlike the case in conventional aqueous electrolytes featuring typical Zn[H2O]62+ solvation, a solvation sheath of Zn2+ with the co-participation of the DMC solvent and OTf− anion is found in the formulated H2O + DMC electrolyte, which contributes to the formation of a robust ZnF2 and ZnCO3-rich interphase on Zn. The resultant Zn anode exhibits a high average CE of Zn plating/stripping (99.8% at an areal capacity of 2.5 mA h cm−2) and dendrite-free cycling over 1000 cycles. Furthermore, the H2O + DMC electrolytes sustain stable operation of RAZBs pairing Zn anodes with diverse cathode materials such as vanadium pentoxide, manganese dioxide, and zinc hexacyanoferrate. Rational electrolyte design with organic solvent additives would promote building better aqueous batteries.

LayerCAM: Exploring Hierarchical Class Activation Maps for Localization

Abstract: The class activation maps are generated from the final convolutional layer of CNN. They can highlight discriminative object regions for the class of interest. These discovered object regions have been widely used for weakly-supervised tasks. However, due to the small spatial resolution of the final convolutional layer, such class activation maps often locate coarse regions of the target objects, limiting the performance of weakly-supervised tasks that need pixel-accurate object locations. Thus, we aim to generate more fine-grained object localization information from the class activation maps to locate the target objects more accurately. In this paper, by rethinking the relationships between the feature maps and their corresponding gradients, we propose a simple yet effective method, called LayerCAM. It can produce reliable class activation maps for different layers of CNN. This property enables us to collect object localization information from coarse (rough spatial localization) to fine (precise fine-grained details) levels. We further integrate them into a high-quality class activation map, where the object-related pixels can be better highlighted. To evaluate the quality of the class activation maps produced by LayerCAM, we apply them to weakly-supervised object localization and semantic segmentation. Experiments demonstrate that the class activation maps generated by our method are more effective and reliable than those by the existing attention methods. The code will be made publicly available.

Underwater Image Enhancement via Medium Transmission-Guided Multi-Color Space Embedding

Abstract: Underwater images suffer from color casts and low contrast due to wavelength- and distance-dependent attenuation and scattering. To solve these two degradation issues, we present an underwater image enhancement network via medium transmission-guided multi-color space embedding, called Ucolor . Concretely, we first propose a multi-color space encoder network, which enriches the diversity of feature representations by incorporating the characteristics of different color spaces into a unified structure. Coupled with an attention mechanism, the most discriminative features extracted from multiple color spaces are adaptively integrated and highlighted. Inspired by underwater imaging physical models, we design a medium transmission (indicating the percentage of the scene radiance reaching the camera)-guided decoder network to enhance the response of network towards quality-degraded regions. As a result, our network can effectively improve the visual quality of underwater images by exploiting multiple color spaces embedding and the advantages of both physical model-based and learning-based methods. Extensive experiments demonstrate that our Ucolor achieves superior performance against state-of-the-art methods in terms of both visual quality and quantitative metrics. The code is publicly available at: https://li-chongyi.github.io/Proj_Ucolor.html .

Almost 100 % Peroxymonosulfate Conversion to Singlet Oxygen on Single-Atom CoN2+2 Sites

Abstract: Single-atom CoN 4 active sites have demonstrated excellent efficiency in peroxymonosulfate activation. However, the identification of CoN 4 active sites and the detailed singlet oxygen generation mechanism in peroxymonosulfate activation still remain ambiguous. In this study, we demonstrated a strategy to regulate the generation of reactive oxygen species by atomically dispersed cobalt anchored on nitrogen-doped carbon. As assisted by experimental and DFT calculations, CoN 2+2 was the definite active sites. Singlet oxygen was the absolutely predominant reactive oxygen species that the proportion was 98.89%. Different from the traditional CoN 4 configuration, the CoN 2+2 active sites transformed the pathway of peroxymonosulfate activation and facilitated the singlet oxygen generation. Spontaneous dissociation of adsorbed peroxymonosulfate on Co single atoms was prevented due to the energy barriers caused by weak positive Co atoms and CoN 2+2 coordination, determining PMS oxidation in non-radical pathway and simultaneous singlet oxygen generation. The generated singlet oxygen showed efficient activity for degradation of several organic pollutants in a broad pH range.

Nitrogen doped g-C3N4 with the extremely narrow band gap for excellent photocatalytic activities under visible light

Abstract: In this work, nitrogen doped g-C3N4 (NCN) with the extremely narrow band gap was prepared and applied for the photodegradation of phenols. Experiments and DFT (the density functional theory) computation identified that N-doping introduced in the g-C3N4 matrix by substituting C atoms. DFT, PL (photoluminescence) spectra, optical property characteristic and PEC (photoelectrochemical) indicated that NCN possess extremely narrow band gap, efficient photogenerated carrier separation and the charge transfer, which enhanced the absorption of visible light and further promoted the photocatalytic activity. As a result, NCN(2:2) showed about twice higher photodegradation efficiencies and 3 times rate constant than pristine g-C3N4. The radical trapping experiments showed that •O2- radical and h+ served as crucial active species during the whole photodegradation reaction. This work can provide a strategy to enhance the photocatalytic activity of photocatalysts via introduce foreign atoms in matrix.

Sandwich-Like Heterostructures of MoS2/Graphene with Enlarged Interlayer Spacing and Enhanced Hydrophilicity as High-Performance Cathodes for Aqueous Zinc-Ion Batteries

Abstract: Layered materials have great potential as cathodes for aqueous zinc-ion batteries (AZIBs) because of their facile 2D Zn2+ transport channels; however, either low capacity or poor cycling stability limits their practical applications. Herein, two classical layered materials are innovatively combined by intercalating graphene into MoS2 gallery, which results in significantly enlarged MoS2 interlayers (from 0.62 to 1.16 nm) and enhanced hydrophilicity. The sandwich-structured MoS2 /graphene nanosheets self-assemble into a flower-like architecture that facilitates Zn-ion diffusion, promotes electrolyte infiltration, and ensures high structural stability. Therefore, this novel MoS2 /graphene nanocomposite exhibits exceptional high-rate capability (285.4 mA h g-1 at 0.05 A g-1 with 141.6 mA h g-1 at 5 A g-1 ) and long-term cycling stability (88.2% capacity retention after 1800 cycles). The superior Zn2+ migration kinetics and desirable pseudocapacitive behaviors are confirmed by electrochemical measurements and density functional theory computations. The energy storage mechanism regarding the highly reversible phase transition between 2H- and 1T-MoS2 upon Zn-ion insertion/extraction is elucidated through ex situ investigations. As a proof of concept, a flexible quasi-solid-state zinc-ion battery employing the MoS2 /graphene cathode demonstrates great stability under different bending conditions. This study paves a new direction for the design and on-going development of 2D materials as high-performance cathodes for AZIBs.

A Coevolutionary Framework for Constrained Multiobjective Optimization Problems

Abstract: Constrained multiobjective optimization problems (CMOPs) are challenging because of the difficulty in handling both multiple objectives and constraints While some evolutionary algorithms have demonstrated high performance on most CMOPs, they exhibit bad convergence or diversity performance on CMOPs with small feasible regions To remedy this issue, this article proposes a coevolutionary framework for constrained multiobjective optimization, which solves a complex CMOP assisted by a simple helper problem The proposed framework evolves one population to solve the original CMOP and evolves another population to solve a helper problem derived from the original one While the two populations are evolved by the same optimizer separately, the assistance in solving the original CMOP is achieved by sharing useful information between the two populations In the experiments, the proposed framework is compared to several state-of-the-art algorithms tailored for CMOPs High competitiveness of the proposed framework is demonstrated by applying it to 47 benchmark CMOPs and the vehicle routing problem with time windows

Highlighting photonics: looking into the next decade

Abstract: Let there be light–to change the world we want to be! Over the past several decades, and ever since the birth of the first laser, mankind has witnessed the development of the science of light, as light-based technologies have revolutionarily changed our lives. Needless to say, photonics has now penetrated into many aspects of science and technology, turning into an important and dynamically changing field of increasing interdisciplinary interest. In this inaugural issue of eLight, we highlight a few emerging trends in photonics that we think are likely to have major impact at least in the upcoming decade, spanning from integrated quantum photonics and quantum computing, through topological/non-Hermitian photonics and topological insulator lasers, to AI-empowered nanophotonics and photonic machine learning. This Perspective is by no means an attempt to summarize all the latest advances in photonics, yet we wish our subjective vision could fuel inspiration and foster excitement in scientific research especially for young researchers who love the science of light.

The trRosetta server for fast and accurate protein structure prediction

Abstract: The trRosetta (transform-restrained Rosetta) server is a web-based platform for fast and accurate protein structure prediction, powered by deep learning and Rosetta. With the input of a protein’s amino acid sequence, a deep neural network is first used to predict the inter-residue geometries, including distance and orientations. The predicted geometries are then transformed as restraints to guide the structure prediction on the basis of direct energy minimization, which is implemented under the framework of Rosetta. The trRosetta server distinguishes itself from other similar structure prediction servers in terms of rapid and accurate de novo structure prediction. As an illustration, trRosetta was applied to two Pfam families with unknown structures, for which the predicted de novo models were estimated to have high accuracy. Nevertheless, to take advantage of homology modeling, homologous templates are used as additional inputs to the network automatically. In general, it takes ~1 h to predict the final structure for a typical protein with ~300 amino acids, using a maximum of 10 CPU cores in parallel in our cluster system. To enable large-scale structure modeling, a downloadable package of trRosetta with open-source codes is available as well. A detailed guidance for using the package is also available in this protocol. The server and the package are available at https://yanglab.nankai.edu.cn/trRosetta/ and https://yanglab.nankai.edu.cn/trRosetta/download/ , respectively. The trRosetta server, a web-based platform for fast and accurate protein structure prediction, is powered by deep learning and Rosetta. This protocol includes procedures for using the web-based server as well as the standalone package.

SpaGCN: Integrating gene expression, spatial location and histology to identify spatial domains and spatially variable genes by graph convolutional network.

Abstract: Recent advances in spatially resolved transcriptomics (SRT) technologies have enabled comprehensive characterization of gene expression patterns in the context of tissue microenvironment. To elucidate spatial gene expression variation, we present SpaGCN, a graph convolutional network approach that integrates gene expression, spatial location and histology in SRT data analysis. Through graph convolution, SpaGCN aggregates gene expression of each spot from its neighboring spots, which enables the identification of spatial domains with coherent expression and histology. The subsequent domain guided differential expression (DE) analysis then detects genes with enriched expression patterns in the identified domains. Analyzing seven SRT datasets using SpaGCN, we show it can detect genes with much more enriched spatial expression patterns than competing methods. Furthermore, genes detected by SpaGCN are transferrable and can be utilized to study spatial variation of gene expression in other datasets. SpaGCN is computationally fast, platform independent, making it a desirable tool for diverse SRT studies.

Learning to Enhance Low-Light Image via Zero-Reference Deep Curve Estimation.

Abstract: This paper presents a novel method, Zero-Reference Deep Curve Estimation (Zero-DCE), which formulates light enhancement as a task of image-specific curve estimation with a deep network. Our method trains a lightweight deep network to estimate pixel-wise and high-order curves for dynamic range adjustment of a given image. The curve estimation is specially designed, considering pixel value range, monotonicity, and differentiability. Zero-DCE is appealing in its relaxed assumption on reference images, i.e., it does not require any paired or unpaired data during training. This is achieved through a set of carefully formulated non-reference losses, which implicitly measure the enhancement quality and drive the learning of the network. Despite its simplicity, it generalizes well to diverse lighting conditions. Our method is efficient as image enhancement can be achieved by a simple nonlinear curve mapping. We further present an accelerated and light version of Zero-DCE, called Zero-DCE++, that takes advantage of a tiny network with just 10K parameters. Zero-DCE++ has a fast inference speed (1000/11 FPS on single GPU/CPU) while keeping the enhancement performance of Zero-DCE. Experiments on various benchmarks demonstrate the advantages of our method over state-of-the-art methods. The potential benefits of our method to face detection in the dark are discussed.

A Review on Metal-Organic Framework-Derived Porous Carbon-Based Novel Microwave Absorption Materials.

Abstract: The development of microwave absorption materials (MAMs) is a considerable important topic because our living space is crowed with electromagnetic wave which threatens human’s health. And MAMs are also used in radar stealth for protecting the weapons from being detected. Many nanomaterials were studied as MAMs, but not all of them have the satisfactory performance. Recently, metal–organic frameworks (MOFs) have attracted tremendous attention owing to their tunable chemical structures, diverse properties, large specific surface area and uniform pore distribution. MOF can transform to porous carbon (PC) which is decorated with metal species at appropriate pyrolysis temperature. However, the loss mechanism of pure MOF-derived PC is often relatively simple. In order to further improve the MA performance, the MOFs coupled with other loss materials are a widely studied method. In this review, we summarize the theories of MA, the progress of different MOF-derived PC‑based MAMs, tunable chemical structures incorporated with dielectric loss or magnetic loss materials. The different MA performance and mechanisms are discussed in detail. Finally, the shortcomings, challenges and perspectives of MOF-derived PC‑based MAMs are also presented. We hope this review could provide a new insight to design and fabricate MOF-derived PC-based MAMs with better fundamental understanding and practical application.

Flexible and efficient perovskite quantum dot solar cells via hybrid interfacial architecture.

Abstract: All-inorganic CsPbI3 perovskite quantum dots have received substantial research interest for photovoltaic applications because of higher efficiency compared to solar cells using other quantum dots materials and the various exciting properties that perovskites have to offer These quantum dot devices also exhibit good mechanical stability amongst various thin-film photovoltaic technologies We demonstrate higher mechanical endurance of quantum dot films compared to bulk thin film and highlight the importance of further research on high-performance and flexible optoelectronic devices using nanoscale grains as an advantage Specifically, we develop a hybrid interfacial architecture consisting of CsPbI3 quantum dot/PCBM heterojunction, enabling an energy cascade for efficient charge transfer and mechanical adhesion The champion CsPbI3 quantum dot solar cell has an efficiency of 151% (stabilized power output of 1461%), which is among the highest report to date Building on this strategy, we further demonstrate a highest efficiency of 123% in flexible quantum dot photovoltaics

Reducing the impact of Auger recombination in quasi-2D perovskite light-emitting diodes.

Abstract: Rapid Auger recombination represents an important challenge faced by quasi-2D perovskites, which induces resulting perovskite light-emitting diodes’ (PeLEDs) efficiency roll-off. In principle, Auger recombination rate is proportional to materials’ exciton binding energy (Eb). Thus, Auger recombination can be suppressed by reducing the corresponding materials’ Eb. Here, a polar molecule, p-fluorophenethylammonium, is employed to generate quasi-2D perovskites with reduced Eb. Recombination kinetics reveal the Auger recombination rate does decrease to one-order-of magnitude lower compared to its PEA+ analogues. After effective passivation, nonradiative recombination is greatly suppressed, which enables resulting films to exhibit outstanding photoluminescence quantum yields in a broad range of excitation density. We herein demonstrate the very efficient PeLEDs with a peak external quantum efficiency of 20.36%. More importantly, devices exhibit a record luminance of 82,480 cd m−2 due to the suppressed efficiency roll-off, which represent one of the brightest visible PeLEDs yet. Designing efficient perovskite light-emitting diodes remains a challenge due to the strong Auger recombination and resulting Joule heating. Here, the authors propose polarizable p-fluorophenethylammonium to generate quasi-2D perovskites with reduced binding energy developing perovskite light-emitting diodes with a peak EQE of 20.36% and a maximum luminance of 82,480 cdm-2.