针对农作物害虫检测中害虫目标被遮挡、体色与环境相近等情况导致的目标检测准确率不高的问题,提出了一种基于RT-DETR的农作物害虫检测算法RT-DETR-SDIC。首先,原主干网络的前两层(S2,S3)引入多样分支残差模块(Diverse Branch Residual ...针对农作物害虫检测中害虫目标被遮挡、体色与环境相近等情况导致的目标检测准确率不高的问题,提出了一种基于RT-DETR的农作物害虫检测算法RT-DETR-SDIC。首先,原主干网络的前两层(S2,S3)引入多样分支残差模块(Diverse Branch Residual Block,DBRB),利用多分支拓扑结构以及不同规模的路径提取多尺度的特征信息,在原主干网络的后两层(S4,S5)引入了结合级联注意力的倒立残差移动模块(Invert Residual Mobile Block with Cascade Group Attention,IRMB_CGA),弥补了原主干网络中长距离语义信息无法直接交互的问题,增强了对环境特征的辨别能力;其次,在特征融合网络中,增加了无参数注意力的空间到深度融合层(Space to Depth Convolution with Attention,SPA)提取细粒度的信息,设计了内容引导融合模块(Context Guide Fusion Module,CGFM)来引导多尺度特征融合。实验结果表明模型RT-DETR-SDIC参数下降了19.6%,计算量下降了9.9%,P_(mA,0.5)上升了6.2%,P_(mA,0.5:0.95)上升了2.6%。展开更多
黄螟Tetramoera schistaceana(Snellen)是甘蔗上的重要钻蛀性害虫。近年来,黄螟种群数量呈增长趋势,已跃升为广西蔗区的优势害虫种群。为明确影响黄螟自然种群发生发展的各项作用因子及其作用大小,2023年和2024年采用田间调查、室内饲...黄螟Tetramoera schistaceana(Snellen)是甘蔗上的重要钻蛀性害虫。近年来,黄螟种群数量呈增长趋势,已跃升为广西蔗区的优势害虫种群。为明确影响黄螟自然种群发生发展的各项作用因子及其作用大小,2023年和2024年采用田间调查、室内饲养观察以及桶栽甘蔗接虫试验相结合的方法,研究了甘蔗苗期黄螟各虫期种群数量变动规律与关键致死因子。构建黄螟自然种群生命表,运用种群数量排除作用控制指数(exclusion index of population control,EIPC)分析不同作用因子对黄螟种群数量的控制作用。结果表明,2023年和2024年黄螟自然种群趋势指数分别达到19.6425和19.3955,幼虫低龄期(1~3龄)的存活率均最低,是种群数量控制的关键期。在黄螟整个生长发育过程中,“捕食及病原菌”因素的种群控制效应最大,赤眼蜂作为卵期的优势天敌,对黄螟自然种群数量调控起关键作用。人工释放赤眼蜂后,2023年和2024年其对黄螟的排除作用控制指数比对照区分别提高1.0540和0.9632,控害效果显著。天敌赤眼蜂对黄螟种群数量有较为明显的控制作用。该结果为黄螟预测预报和综合防治提供坚实的理论基础。展开更多
YOLO(You Only Look Once)是一种流行的目标检测模型,该模型利用单一神经网络架构即可对图像中的目标进行位置和类别的预测,极大地提高了检测的效率。将YOLO模型应用于植物虫害的检测,不仅可以加快识别速度,还可以提升识别精度。文章阐...YOLO(You Only Look Once)是一种流行的目标检测模型,该模型利用单一神经网络架构即可对图像中的目标进行位置和类别的预测,极大地提高了检测的效率。将YOLO模型应用于植物虫害的检测,不仅可以加快识别速度,还可以提升识别精度。文章阐述了YOLO模型目前在植物虫害检测领域的应用现状,介绍了该模型的优势与面临的挑战,并对其在未来发展进程中的改进策略与方向进行了展望。展开更多
The red imported fire ant,Solenopsis invicta Buren,is a highly invasive eusocial insect pest that threatens native biodiversity,agriculture,and human health.The innate immune system and intricate social immune respons...The red imported fire ant,Solenopsis invicta Buren,is a highly invasive eusocial insect pest that threatens native biodiversity,agriculture,and human health.The innate immune system and intricate social immune responses of S.invicta pose challenges to the development of effective control strategies.Micro RNAs(mi RNAs)play critical roles in the post-transcriptional regulation of gene expression,which influences various biological processes,including immunity and host-pathogen interactions.While the mi RNA-mediated response of insects to pathogens has been extensively studied in solitary insects,little is known about the innate immune responses of individual members within a colony.To address this gap,we constructed small RNA libraries from Metarhizium anisopliae-infected S.invicta workers and investigated the temporal dynamics of mi RNA-mediated immune responses to the entomopathogen.Several differentially expressed mi RNAs were identified,and they were found to regulate genes involved in the Toll,IMD,and melanization immune pathways.Quantitative real-time PCR(q RT-PCR)was employed to analyze the spatiotemporal dynamics of key mi RNAs/target genes,specifically mi R-71/Mod SP1-Relish and mi R-7/Lysozyme2-Serine protease7.A dual luciferase assay(in vitro)was performed to validate the interactions between mi RNAs and their target genes.Overexpression of mi R-71 and mi R-7(via mi RNA mimics)efficiently suppressed their target genes,impaired the antifungal immune response of S.invicta and increased the susceptibility to M.anisopliae infection compared to controls.Furthermore,RNA interference-based gene silencing elucidated the roles of these immune genes in regulating fungal susceptibility,thus providing vital clues for developing virulent and effective mycoinsecticides using modern genetic engineering tools.展开更多
文摘针对农作物害虫检测中害虫目标被遮挡、体色与环境相近等情况导致的目标检测准确率不高的问题,提出了一种基于RT-DETR的农作物害虫检测算法RT-DETR-SDIC。首先,原主干网络的前两层(S2,S3)引入多样分支残差模块(Diverse Branch Residual Block,DBRB),利用多分支拓扑结构以及不同规模的路径提取多尺度的特征信息,在原主干网络的后两层(S4,S5)引入了结合级联注意力的倒立残差移动模块(Invert Residual Mobile Block with Cascade Group Attention,IRMB_CGA),弥补了原主干网络中长距离语义信息无法直接交互的问题,增强了对环境特征的辨别能力;其次,在特征融合网络中,增加了无参数注意力的空间到深度融合层(Space to Depth Convolution with Attention,SPA)提取细粒度的信息,设计了内容引导融合模块(Context Guide Fusion Module,CGFM)来引导多尺度特征融合。实验结果表明模型RT-DETR-SDIC参数下降了19.6%,计算量下降了9.9%,P_(mA,0.5)上升了6.2%,P_(mA,0.5:0.95)上升了2.6%。
文摘黄螟Tetramoera schistaceana(Snellen)是甘蔗上的重要钻蛀性害虫。近年来,黄螟种群数量呈增长趋势,已跃升为广西蔗区的优势害虫种群。为明确影响黄螟自然种群发生发展的各项作用因子及其作用大小,2023年和2024年采用田间调查、室内饲养观察以及桶栽甘蔗接虫试验相结合的方法,研究了甘蔗苗期黄螟各虫期种群数量变动规律与关键致死因子。构建黄螟自然种群生命表,运用种群数量排除作用控制指数(exclusion index of population control,EIPC)分析不同作用因子对黄螟种群数量的控制作用。结果表明,2023年和2024年黄螟自然种群趋势指数分别达到19.6425和19.3955,幼虫低龄期(1~3龄)的存活率均最低,是种群数量控制的关键期。在黄螟整个生长发育过程中,“捕食及病原菌”因素的种群控制效应最大,赤眼蜂作为卵期的优势天敌,对黄螟自然种群数量调控起关键作用。人工释放赤眼蜂后,2023年和2024年其对黄螟的排除作用控制指数比对照区分别提高1.0540和0.9632,控害效果显著。天敌赤眼蜂对黄螟种群数量有较为明显的控制作用。该结果为黄螟预测预报和综合防治提供坚实的理论基础。
文摘YOLO(You Only Look Once)是一种流行的目标检测模型,该模型利用单一神经网络架构即可对图像中的目标进行位置和类别的预测,极大地提高了检测的效率。将YOLO模型应用于植物虫害的检测,不仅可以加快识别速度,还可以提升识别精度。文章阐述了YOLO模型目前在植物虫害检测领域的应用现状,介绍了该模型的优势与面临的挑战,并对其在未来发展进程中的改进策略与方向进行了展望。
基金supported by grants from the National Natural Science Foundation of China(32172498 and W2433052)the National Key R&D Program of China(2021YFD1000500)the Natural Science Foundation of Guangdong,China(2023A1515010305)。
文摘The red imported fire ant,Solenopsis invicta Buren,is a highly invasive eusocial insect pest that threatens native biodiversity,agriculture,and human health.The innate immune system and intricate social immune responses of S.invicta pose challenges to the development of effective control strategies.Micro RNAs(mi RNAs)play critical roles in the post-transcriptional regulation of gene expression,which influences various biological processes,including immunity and host-pathogen interactions.While the mi RNA-mediated response of insects to pathogens has been extensively studied in solitary insects,little is known about the innate immune responses of individual members within a colony.To address this gap,we constructed small RNA libraries from Metarhizium anisopliae-infected S.invicta workers and investigated the temporal dynamics of mi RNA-mediated immune responses to the entomopathogen.Several differentially expressed mi RNAs were identified,and they were found to regulate genes involved in the Toll,IMD,and melanization immune pathways.Quantitative real-time PCR(q RT-PCR)was employed to analyze the spatiotemporal dynamics of key mi RNAs/target genes,specifically mi R-71/Mod SP1-Relish and mi R-7/Lysozyme2-Serine protease7.A dual luciferase assay(in vitro)was performed to validate the interactions between mi RNAs and their target genes.Overexpression of mi R-71 and mi R-7(via mi RNA mimics)efficiently suppressed their target genes,impaired the antifungal immune response of S.invicta and increased the susceptibility to M.anisopliae infection compared to controls.Furthermore,RNA interference-based gene silencing elucidated the roles of these immune genes in regulating fungal susceptibility,thus providing vital clues for developing virulent and effective mycoinsecticides using modern genetic engineering tools.
