Lightweight Method for Plant Disease Identification Using Deep Learning

In the deep learning approach for identifying plant diseases,the high complexity of the network model,the large number of parameters,and great computational effort make it challenging to deploy the model on terminal devices with limited computational resources.In this study,a lightweight method for plant diseases identification that is an improved version of the ShuffleNetV2 model is proposed.In the proposed model,the depthwise convolution in the basic module of ShuffleNetV2 is replaced with mixed depthwise convolution to capture crop pest images with different resolutions;the efficient channel attention module is added into the ShuffleNetV2 model network structure to enhance the channel features;and the ReLU activation function is replaced with the ReLU6 activation function to prevent the gen-eration of large gradients.Experiments are conducted on the public dataset PlantVillage.The results show that the proposed model achieves an accuracy of 99.43%,which is an improvement of 0.6 percentage points compared to the ShuffleNetV2 model.Compared to lightweight network models,such as MobileNetV2,MobileNetV3,EfficientNet,and EfficientNetV2,and classical convolutional neural network models,such as ResNet34,ResNet50,and ResNet101,the proposed model has fewer parameters and higher recognition accuracy,which provides guidance for deploying crop pest identification methods on resource-constrained devices,including mobile terminals.

High power tunable Raman fiber laser at 1.2μm waveband

Development of a high power fiber laser at special waveband,which is difficult to achieve by conventional rare-earth-doped fibers,is a significant challenge.One of the most common methods for achieving lasing at special wavelength is Raman conversion.Phosphorus-doped fiber(PDF),due to the phosphorus-related large frequency shift Raman peak at 40 THz,is a great choice for large frequency shift Raman conversion.Here,by adopting 150 m large mode area triple-clad PDF as Raman gain medium,and a novel wavelength-selective feedback mechanism to suppress the silica-related Raman emission,we build a high power cladding-pumped Raman fiber laser at 1.2μm waveband.A Raman signal with power up to 735.8 W at 1252.7 nm is obtained.To the best of our knowledge,this is the highest output power ever reported for fiber lasers at 1.2μm waveband.Moreover,by tuning the wavelength of the pump source,a tunable Raman output of more than 450 W over a wavelength range of 1240.6–1252.7 nm is demonstrated.This work proves PDF’s advantage in high power large frequency shift Raman conversion with a cladding pump scheme,thus providing a good solution for a high power laser source at special waveband.

High power cladding-pumped low quantum defect Raman fiber amplifier

Heat generated by the quantum defect(QD)in optically pumped lasers can result in detrimental effects such as mode instability,frequency noise,and even catastrophic damage.Previously,we demonstrated that boson-peakbased Raman fiber lasers have great potential in low QD laser generation.But their power scalability and heat load characteristics have yet to be investigated.Here,we demonstrate a boson-peak-based Raman fiber amplifier(RFA)with 815 W output power and a QD of 1.3%.The low heat generation characteristics of this low QD RFA are demonstrated.Both experimental and simulation results show that at this power level,the heat load of the low QD RFA is significantly lower than that of the conventional RFA with a QD of 4.8%.Thanks to its low heat generation characteristics,the proposed phosphosilicate-fiber-based low QD RFA provides an effective solution for the intractable thermal issue in optically pumped lasers,which is of significance in reducing the laser’s noise,improving the laser’s stability and safety,and solving the challenge of heat removing.

2kW random fiber laser based on hybrid Yb-Raman gain [Invited]

High-power operation is one of the most important research topics surrounding random fiber lasers(RDFLs).Here we optimized the cavity structure and proposed a new scheme based on hybrid gain to address the issue of high-power backward light in traditional kilowatt-level RDFLs.Consequently,a record power of 1972 W was achieved while the maximum backward leaked power only reached 0.12 W.The conversion efficiency relative to the laser diode pump power was 68.4%,and the highest spectral purity of the random lasing reached 98.1%.This work may provide a reference for high-power RDFLs,Raman fiber lasers,and long-wavelength Yb-doped fiber lasers.

Cladding-pumped Raman fiber laser with 0.78%quantum defect enabled by phosphorus-doped fiber

The quantum defect(QD)is an important issue that demands prompt attention in high-power fiber lasers.A large QD may aggravate the thermal load in the laser,which would impact the frequency,amplitude noise and mode stability,and threaten the security of the high-power laser system.Here,we propose and demonstrate a cladding-pumped Raman fiber laser(RFL)with QD of less than 1%.Using the Raman gain of the boson peak in a phosphorus-doped fiber to enable the cladding pump,the QD is reduced to as low as 0.78%with a 23.7 W output power.To our knowledge,this is the lowest QD ever reported in a cladding-pumped RFL.Furthermore,the output power can be scaled to 47.7 W with a QD of 1.29%.This work not only offers a preliminary platform for the realization of high-power low-QD fiber lasers,but also proves the great potential of low-QD fiber lasers in power scaling.

Revealing the dynamics of intensity fluctuation transfer in a random Raman fiber laser

Temporal intensity fluctuation is one of the inherent features of fiber lasers.When utilizing the fiber lasers to pump a random Raman fiber laser(RRFL),the intensity fluctuation transfer from the pump to the random lasing could affect the output performance significantly.In this paper,we comprehensively compared the spectral,temporal,and power characteristics of an RRFL pumped by two different fiber lasers—a temporally unstable fiber oscillator and a temporally stable amplified spontaneous emission(ASE)source.Owing to less impact of the intensity fluctuation transfer,the ASE source-pumped RRFL shows∼45.3%higher maximum output power,higher spectral purity(>99.9%)and optical signal-to-noise ratio(>40dB),weaker spectral broadening,and more stable temporal behavior compared to the fiber oscillator-pumped RRFL.Furthermore,based on the temporal-spatial-coupled Raman equations and the generalized nonlinear Schrödinger equations,we numerically revealed the impact of the pump intensity fluctuations on the output characteristics of RRFLs,and found that the temporal walk-off effect played an important role in the dynamics of intensity fluctuation transfer.This work may provide a reference for designing and implementing high-performance RRFLs and promote their practicability in sensing,telecommunications,and high-power applications.

Multi-wavelength random fiber laser with a spectral-flexible characteristic

In past decades,multi-wavelength lasers have attracted much attention due to their wide applications in many fields.In this paper,we demonstrate a multi-wavelength random fiber laser with customizable spectra enabled by an acousto–optic tunable filter.The operating wavelength range can be tuned from 1114.5 to 1132.5 nm with a maximal output power of 5.55 W,and spectral channel tuning can also be realized with a maximal number of five.The effect of gain competition and the interaction between Raman gain and insertion loss are also analyzed.Furthermore,the output spectra can be ordered by radiating appropriate radio frequency signals to the acousto–optic tunable filter.This work may provide a reference for agile shape spectrum generation and promote multi-wavelength random fiber laser practicability in sensing,telecommunications,and precise spectroscopy.

Vortex random fiber laser with controllable orbital angular momentum mode

In this paper, we propose and experimentally demonstrate a vortex random fiber laser(RFL) with a controllable orbital angular momentum(OAM) mode. The topological charge of the vortex RFL can range from-50 to 50 with nearly watt-level output power. A triangular toroidal interferometer is constructed to verify the spiral phase structure of the generated vortex random laser with a special coherence property. Vortex RFLs with fractional topological charge are also performed in this work. As the first demonstration of a vortex RFL with a controllable OAM mode(to the best of our knowledge), this work may not only offer a valuable reference on temporal modulation of a vortex beam and optical field control of an RFL but also provide a potential vortex laser source for applications in imaging, sensing, and communication.

Notes for genera:Ascomycota

Knowledge of the relationships and thus the classification of fungi,has developed rapidly with increasingly widespread use of molecular techniques,over the past 10–15 years,and continues to accelerate.Several genera have been found to be polyphyletic,and their generic concepts have subsequently been emended.New names have thus been introduced for species which are phylogenetically distinct from the type species of particular genera.The ending of the separate naming of morphs of the same species in 2011,has also caused changes in fungal generic names.In order to facilitate access to all important changes,it was desirable to compile these in a single document.The present article provides a list of generic names of Ascomycota(approximately 6500 accepted names published to the end of 2016),including those which are lichen-forming.Notes and summaries of the changes since the last edition of‘Ainsworth&Bisby’s Dictionary of the Fungi’in 2008 are provided.The notes include the number of accepted species,classification,type species(with location of the type material),culture availability,life-styles,distribution,and selected publications that have appeared since 2008.This work is intended to provide the foundation for updating the ascomycete component of the"Without prejudice list of generic names of Fungi"published in 2013,which will be developed into a list of protected generic names.This will be subjected to the XIXth International Botanical Congress in Shenzhen in July 2017 agreeing to a modification in the rules relating to protected lists,and scrutiny by procedures determined by the Nomenclature Committee for Fungi(NCF).The previously invalidly published generic names Barriopsis,Collophora(as Collophorina),Cryomyces,Dematiopleospora,Heterospora(as Heterosporicola),Lithophila,Palmomyces(as Palmaria)and Saxomyces are validated,as are two previously invalid family names,Bartaliniaceae and Wiesneriomycetaceae.Four species of Lalaria,which were invalidly published are transferred to Taphrina and validated as new combinations.Catenomycopsis Tibell&Constant.is reduced under Chaenothecopsis Vain.,while Dichomera Cooke is reduced under Botryosphaeria Ces.&De Not.(Art.59).

Origin and evolution of green plants in the light of key evolutionary events

Green plants(Viridiplantae)are ancient photosynthetic organisms that thrive both in aquatic and terrestrial ecosystems,greatly contributing to the changes in global climates and ecosystems.Significant progress has been made toward understanding the origin and evolution of green plants,and plant biologists have arrived at the consensus that green plants first originated in marine deep-water environments and later colonized fresh water and dry land.The origin of green plants,colonization of land by plants and rapid radiation of angiosperms are three key evolutionary events during the long history of green plants.However,the comprehensive understanding of evolutionary features and molecular innovations that enabled green plants to adapt to complex and changeable environments are still limited.Here,we review current knowledge of phylogenetic relationships and divergence times of green plants,and discuss key morphological innovations and distinct drivers in the evolution of green plants.Ultimately,we highlight fundamental questions to advance our understanding of the phenotypic novelty,environmental adaptation,and domestication of green plants.