An Intelligence Decision Support System for Diagnosis and Management of Grain and Cotton Crop Pests in Xinjiang

Xinjiang Uygur Autonomous Region is a scarcely populated area in China and technicians for plant protection are extremely deficient.The occurrence areas of insect pests in grain and cotton crops have been increasing year by year, causing serious economic losses. Aiming for several main grain and economic crops of Xinjiang(cotton, corn and wheat), an intelligence decision support system for diagnosis and management of grain and cotton crop pests in Xinjiang was designed and developed, which was based on android platform and windows system architecture. APP application program of smart phones was used as an implementation form. The intelligence decision support system will help plant protection personnel and farmers to solve local pest recognition and prevention control problem at the grassroots level in Xinjiang remote regions.

Illustrations with Real Examples of Using Ecological Regulation Strategies Against Crop Pests in China

In this paper, a series of examples from China, including those using ecological regulation strategies successfully to keep crop pests under control and those making more damages of crop pests caused by wrong cropping systems, crop distributions or extensive cultivation, were given and analyzed. From these, two pieces of experience were enlightened as follows: the first, the reforms of cropping systems and crop distributions must be beneficial to pest control or at least not increase the damage of crop pests; the second, we should further clarify the relationship between population dynamics of pests and cropping system/crop distribution, the basis for correctly laying down the ecological regulation systems.

RESEARCH AND APPLICATION OF CROP PEST MONITORING AND EARLY WARNING TECHNOLOGY IN CHINA

The importance of food security,especially in combating the problem of acute hunger,has been underscored as a key component of sustainable development.Considering the major challenge of rapidly increasing demands for both food security and safety,the management and control of major pests is urged to secure supplies of major agricultural products.However,owing to global climate change,biological invasion(e.g.,fall armyworm),decreasing agricultural biodiversity,and other factors,a wide range of crop pest outbreaks are becoming more frequent and serious,making China,one of the world’s largest country in terms of agricultural production,one of the primary victims of crop yield loss and the largest pesticide consumer in the world.Nevertheless,the use of science and technology in monitoring and early warning of major crop pests provides better pest management and acts as a fundamental part of an integrated plant protection strategy to achieve the goal of sustainable development of agriculture.This review summarizes the most fundamental information on pest monitoring and early warning in China by documenting the developmental history of research and application,Chinese laws and regulations related to plant protection,and the National Monitoring and Early Warning System,with the purpose of presenting the Chinese model as an example of how to promote regional management of crop pests,especially of cross border pests such as fall armyworm and locust,by international cooperation across pest-related countries.

Crop diversity and pest management in sustainable agriculture

Large-scale crop monocultures facilitate the proliferation and increasing prevalence of diseases and pest insects.Many studies highlight the impacts of plant diversification upon pathogens,and the population dynamics of insects and beneficial organism in agricultural ecosystems.These studies provide evidence that habitat manipulation techniques such as intercropping,relay,and rotation can significantly improve disease and pest management.This review introduces the concept of crop diversity,considers recent insights and mechanisms underlying crop diversity,and discusses its potential for improving sustainable agricultural practices.Recently,the phytobiomes resulting from increased crop diversity are increasingly recognized for their contribution to disease and pest control.Further,understanding the interactions between pathogens or pests with their host phytobiome may lead to novel options for the prevention of pests.Recent advances in the agricultural systems include:(i)a better understanding of the mechanisms of interactions between crop species and genotypes;(ii)ecological progress including a better understanding of the context-dependency of those interactions;and(iii)the role of microtopographic variation in agricultural systems for priming basal resistance to multiple pests and pathogens by intercropped crops.We also highlight recent progress in China and the potential options for habitat management and design that enhance the ecological role of biodiversity in agroecosystems.

Use of biological control against arthropod pests in Canadian greenhouse crop production

Greenhouse horticultural production currently represents an important and growing sector of Canada's food and plant production systems. Since 2006,the value of greenhouse vegetable crops in Canada exceeds that of field grown crops,signaling an important shift in the way food is cultivated in the country. While many factors have contributed to this change,a major area of innovation includes the discoveries and advances made in the development of commercial greenhouse production systems as well as the integration of biological control strategies for sustainable pest management. With this focus,this review offers a brief overview of the Canadian greenhouse industry,including a descriptive list of commonly used biological control organisms,as well as the role Canadian research has played in the development of these agents. We also address the threats that Canadian greenhouse producers face by invasive pests and the complications these have created for the commercialization of novel biological control agents. This information may serve as a guide for the development of parallel technologies and tools in other parts of the world where greenhouse production is expanding.

Natural Enemies and Pest Control in Field-Grown Crop in Southern Senegal

Crop pests are a permanent threat to horticultural production. Several control methods are recommended for their control, including biological control. This method based on the use of natural enemies is not well established. The objective of this study was to assess the diversity and abundance of natural enemies of crop pests. A survey of crops associated with pests was carried out on 144 fields in southern Senegal to measure the diversity and abundance of natural enemies of various crop pests, determine the native parasitoid complex and assess parasitism rates. Ecological indices were calculated to assess the diversity of natural enemies. A total of 25 natural enemy species were identified, including 15 parasitoids and 10 predators. Predatory Hemiptera and Parasitoid Hymenoptera were high biodiversity. The parasitism rates vary between 0 and 50% depending on the abundance of pests and parasitoid species. Knowledge of the diversity of natural enemies and the understanding of trophic interactions with pests are important in the development of biological crop protection in order to preserve resilience in agroecosystems.

基于PEST的土壤-作物系统模型参数优化及灵敏度分析

农业生产管理系统模型输入参数多,参数率定过程十分耗时费力,大大限制了其推广应用。该研究以华北平原2 a的冬小麦-夏玉米田间试验观测数据为基础,使用PEST(parameter estimation)参数自动优化工具对土壤-作物-大气系统水热碳氮过程藕合模型(soil water heat carbon and nitrogen simulator,WHCNS)的土壤水力学参数、氮素转化参数和作物遗传参数进行自动寻优,同时计算分析模型参数的相对综合敏感度,并将优化结果与土壤实测水力学参数和试错法的模拟结果进行比较。参数敏感度分析结果表明,18个模型参数的相对综合敏感度较高,其中土壤水力学参数普遍具有较高的敏感度,以饱和含水率敏感度最高;作物参数中,作物生长发育总积温和最大比叶面积具有较高的综合敏感度;而氮素转化参数的敏感度远低于土壤水力学参数和作物参数。评价模型模拟效果的统计性指标(均方根误差、模型效率系数和一致性指数)表明,PEST法比实测水力学参数的模拟精度有所提高,其中土壤含水率、土壤硝态氮含量、作物产量和叶面积指数的均方根误差分别降低了61.8%、23.5%、73.6%和23.3%。同时PEST法比试错法对土壤水分和作物产量的模拟精度也有较大提高,但对土壤氮素和叶面积指数的模拟精度提高不明显。由于该方法大大节约了模型校准时间,在较短的时间内获得了明显高于试错法的模拟精度,因此PEST软件在WHCNS模型参数自动优化中是一个值得推广的工具。

基于SWOT-PEST分析的中国热带作物产业发展策略研究

热带作物产业是中国农业的重要组成部分,对保障国家资源安全、乡村振兴、民族团结和服务外交等方面具有重要意义。运用SWOT-PEST方法,结合中国热带作物产业发展现状,从政治、经济、社会、技术4个方面对中国热带作物产业的内部优势、劣势以及面临的外部机遇、挑战进行分析,并构建了分析矩阵,研究表明中国热带作物产业拥有政府重视程度高、自然资源丰富、区位优、较好的产业发展基础和科技力量储备等内部优势,同时也存在扶持力度相对不足、经营水平发展较为滞后、资源环境约束和生产成本的压力较大和科技进步贡献率较低等自身劣势。建议综合运用中国热带作物产业发展的增长型、扭转型、多元化和防御型4种策略,以“一带一路”倡议、乡村振兴战略和持续对外开放政策为契机,利用好国内市场需求旺盛、多元化需求不断增长、新技术应用力度加大的发展机遇,在充分挖掘和发挥内部优势的基础上,通过呼吁政府进一步加强对热带作物产业的政策支持力度、完善热带作物产业经营体系建设、推动热带作物企业“走出去”、科技创新驱动产业延链升级等方式补齐自身短板,减少或避免国际热带农产品市场对国内的冲击、供给安全的不确定性以及科技创新供给不足的威胁,全面提高中国热带作物产业的国际竞争力。

Estimation of the potential geographical distribution of a new potato pest(Schrankia costaestrigalis) in China under climate change

Global food security is threatened by the impacts of the spread of crop pests and changes in the complex interactions between crops and pests under climate change.Schrankia costaestrigalis is a newly-reported potato pest in southern China.Early-warning monitoring of this insect pest could protect domestic agriculture as it has already caused regional yield reduction and/or quality decline in potato production.Our research aimed to confirm the potential geographical distributions(PGDs)of S.costaestrigalis in China under different climate scenarios using an optimal MaxEnt model,and to provide baseline data for preventing agricultural damage by S.costaestrigalis.Our findings indicated that the accuracy of the optimal MaxEnt model was better than the default-setting model,and the minimum temperature of the coldest month,precipitation of the driest month,precipitation of the coldest quarter,and the human influence index were the variables significantly affecting the PGDs of S.costaestrigalis.The highly-and moderately-suitable habitats of S.costaestrigalis were mainly located in eastern and southern China.The PGDs of S.costaestrigalis in China will decrease under climate change.The conversion of the highly-to moderately-suitable habitat will also be significant under climate change.The centroid of the suitable habitat area of S.costaestrigalis under the current climate showed a general tendency to move northeast and to the middle-high latitudes in the 2030s.The agricultural practice of plastic film mulching in potato fields will provide a favorable microclimate for S.costaestrigalis in the suitable areas.More attention should be paid to the early warning and monitoring of S.costaestrigalis in order to prevent its further spread in the main areas in China’s winter potato planting regions.

一种轻量级YOLOv5S农作物虫害目标检测模型

[目的]本文提出一种轻量级YOLOv5S农作物虫害目标检测模型以解决在样本数量不足的情况下农作物虫害识别的问题。[方法]利用Ghost技术将2个Ghost Bottle Block线性特征提取模块封装为1个GB模块,代替YOLOv5S中前7个CBL、CSP、SPP非线性特征提取模块,从而约简了YOLOv5S的网络参数,减轻了网络体量。[结果]在保证虫害检测效果的前提下降低网络对计算硬件与训练样本的依赖。为了验证模型的有效性,对水稻、玉米、棉花、马铃薯、苜蓿5种农作物的12类虫害进行识别与定位,多类别平均精度(mAP)为91.31%,比YOLOv5S模型高出2.56百分点。[结论]通过与SSD、Faster-RCNN、YOLOv5S模型对比发现,本文提出的模型在mAP、F1-score、精确率(Precision)、召回率(Recall)4个评价指标均占优势,尤其在小目标、密集目标、与背景相似目标的检测方面表现突出。

基于渐进式生成对抗网络的农作物病虫害细粒度分类

随着深度学习应用的普及和飞速发展,基于深度学习的图像识别方法广泛应用于农作物病虫害领域,但大部分的神经网络重视识别准确率的提高,却忽略神经网络庞大的参数计算量。为解决这个问题,基于渐进式生成对抗网络判别器模型和卷积注意力模块,提出一种改进的渐进式生成对抗网络判别器CPDM网络模型对农作物病虫害进行识别。通过对渐进式生成对抗网络判别器网络结构的调整,采用均衡学习率、像素级特征向量归一化和卷积注意力模块增强CPDM网络模型的特征提取能力,提高对真实图片的识别准确率。试验在PlantVillage数据集上进行,将该模型与VGG16、VGG19和ResNet18进行比较,得到TOP-1准确率分别为99.06%、96.50%、96.65%、98.86%,分别提高2.56%、2.41%、0.2%,且参数量仅为8.2 M。试验证明提出的CPDM网络模型满足在保证分类准确率的基础上,有效控制神经网络参数计算量的目的。

计算机视觉技术在农作物病虫害检测中的运用研究

现代农业发展更趋集约化、信息化,这为计算机视觉技术的运用提供了空间。以计算机视觉技术在农作物病虫害检测中的运用优势为切入点,在此基础上分析其具体运用方法,包括建设信息作业系统、强调信息采集、运用可视化技术、重视信息复用等内容。最后就该技术在农作物病虫害检测中的未来运用进行展望,一方面客观呈现技术优势、特点,另一方面为其发挥价值提供必要参考。

基于CNN-Transformer的农作物病虫害知识问答意图识别与槽位填充联合模型

意图识别与槽位填充是农作物病虫害知识问答中问题理解的两个重要任务。在已有面向农业领域的研究中,上述任务仍被视为两个完全独立的子任务,并且未充分利用意图识别与槽位填充的语义信息。为此,该研究提出一种基于CNN-Transformer的意图识别与槽位填充联合模型(CDPCT-IDSF)。该模型根据农作物病虫害文本语义复杂设计CNN网络与多层Transformer结合强调局部的有用信息以缓解语义缺失问题;然后在Transformer解码器中引入对齐保证输入与输出一对一关系以提高识别正确槽位标签的能力。此外,进一步构建了包含20个意图类别、12个槽位类别和11 242条标注样本的农业病虫害知识问答数据集进行对比试验,CDPCT-IDSF模型在该语料库上的槽位填充F1值为94.36%,意图识别精度为92.99%,整体识别精度为87.23%,优于其他对比模型,结果证明了所提模型在农作物病虫害意图识别与槽位填充任务上的有效性,可为面向农作物病虫害的知识问答研究提供理论支撑。

轻量化YOLO模型在农作物微小病虫害检测中的应用研究

针对农作物病虫害检测中早期微小病虫害变化目标识别准确率不高的问题,提出一种轻量化农作物微小病虫害检测算法YOLO-MobileNet-CBAM。该算法采用MobileNetV3轻量级卷积模块替换YOLOv5s的主干提取网络来降低参数运算量,并引入CBAM注意力机制从通道和空间两个维度对重要的特征提取进行强化,有效增强小目标的聚焦准确度。卷积模块中用H-SiLU代替原模型的SiLU激活函数提高训练速度,避免梯度消失问题。通过将SIoU函数代替原模型中的GIoU函数计算预测框回归损失,将形状损失计算在内,进一步提高小目标定位精度。通过特征金字塔输出4个不同尺度的检测头识别大面积病害、微小病害及虫害目标,增加微小目标的检测精度。结果表明,YOLO-MobileNet-CBAM在微小病虫害目标检测任务中精确率达92.38%,召回率达90.24%,平均精度大于90%。实现模型轻量化,同时有效提高检测精确度,为手持式终端检测应用提供技术支持。

基于多尺度数据集的虫害检测模型

在复杂和密集尺度的农业检测任务中基于Transformer的DETR类模型逐渐崭露头角,但现有的传统量化分析难以深入探究检测模型的优化策略和机制的有效性。为了满足密集小型昆虫的检测和计数的需求,该研究构建了真实场景采集的包含稻飞虱、蚜虫和麦蜘蛛3类小型害虫的多尺度虫害数据集(multi-scale tiny crowded planthopper-aphid-wheat mite pest detection,MTC-PAWPD),提出基于黏连率和相对尺寸的数据划分方式,将虫害数据集分为密集分布、大尺度、中小尺度和超小尺度4类场景。通过分析DINO模型在特定场景下的性能表现,设计详尽的消融试验方案,以特征与其关键组件查询的可视化试验辅助,验证虫害检测任务性能提高与各个模块之间尤其是与查询相关模块的联系。经试验,模型在MTC-PAWPD测试集上的mAP@50达70.0%,与其他主流模型相比提升2.3个百分点。在4类场景中mAP@50分别达42.5%、79.4%、75.7%和62.4%,在多尺度、密集任务中表现出较强的检测性能。通过针对自身模块的消融和可视化试验证明模型性能提升与Transformer部分的优化模块相关性更大。试验说明基于DINO的虫害检测模型在田间复杂背景下多尺度的复杂虫害检测任务中具有强大的泛化能力和实用价值。

基于卷积神经网络的农作物病虫害检测研究进展

农作物病虫害是全球农业生产的严重威胁之一,易造成巨大的经济损失。引入机器视觉和机器学习方法进行农作物病虫害检测,不仅可以提高病虫害检测的效率,而且有助于及时采取防治措施,降低损失。卷积神经网络(CNN)作为深度学习的代表技术之一,在计算机视觉领域的图像识别、物体识别等方面应用广泛,在农作物病虫害检测方面也取得了一些成果。本文概述了基于CNN检测农作物病虫害的技术要点、发展历程,综述了该技术的主要研究方向与进展,总结了目前研究中存在的主要问题并提出相应的解决策略,旨在为CNN在农业上的应用提供理论依据,并为农业生产管理的智能化提供技术支撑。

基于YOLO算法的农作物病虫害识别研究综述

农作物病虫害是农业生产管理的关键,为及时防控病虫害,人们通过各种技术手段识别和监测病虫害。本文通过介绍目标检测算法YOLO (You Only Look Once)的发展历程及其在农作物病虫害识别中的应用,着重分析了YOLO算法在提高农作物病虫害识别准确度和缩短识别时间的优势,以期为农业生产提供科学指导。

基于知识图谱的农作物病虫害问答系统研究

在大数据时代的推动下,智能问答系统在各个领域得到广泛应用,为用户提供高效的答案,相比传统的文本知识收集和网络搜索引擎检索,具有明显优势。随着知识图谱技术的快速发展,智能问答系统迎来了新的发展阶段。本研究响应了智慧农业的需求,构建了一个基于知识图谱的农作物病虫害智能问答系统,旨在为用户提供与农作物病虫害相关的问答服务。主要工作包括:(1)农作物病虫害数据获取:利用分布式爬虫框架爬取农作物病虫害相关网页数据,并进行数据清理、分析、结构化等预处理操作。(2)知识图谱构建:对数据进行分析后,定义知识本体的实体类别与关系类别,完成知识图谱的模式层构建。利用基于规则的三元组模板对半结构化文本进行实体提取,构建数据层,并将三元组存储至Neo4j图数据库。(3)问答算法设计:利用BERT-BiLSTM-CRF模型进行问句实体识别,BERT-RNN模型进行问题分类。匹配模板后通过Cypher语句进行查询,将答案处理为自然语言形式并返回。(4)问答系统实现与可视化:结合农作物病虫害知识图谱与问答算法,使用Flask框架和多种web技术实现用户提问、实体识别、知识检索和答案返回等功能。试验结果表明,实体识别与问题分类模型的准确率(Precision)、召回率(Recall)和F1分别达到了93.22%、92.69%、92.21%和94.37%、93.92%、92.66%。与其他搜索途径相比,问答系统展现出了较高的准确性和稳定性。这项研究为农业信息化提供了一种智能化的解决方案,为用户获取农作物病虫害知识提供了新途径。

一类具有群体防御行为的庄稼-害虫-天敌模型的动力学分析

研究一类具有群体防御行为的庄稼-害虫-天敌模型的动力学性质.首先,给出了正平衡点的存在条件以及平衡点局部渐近稳定的条件;其次,利用规范型理论和中心流型定理系统分析了Hopf分支的相关性质,包括存在性、稳定性和分支方向等;最后,通过数值模拟验证理论分析的结果.

基于改进YOLOv7的农作物虫害识别

为提高农作物虫害检测的准确率和效率,提出了一种基于YOLOv7的农作物虫害识别模型。首先,使用信息聚集−分发机制改进YOLOv7的特征融合模块,增强了不同层级之间的特征融合能力;其次,将损失函数替换为最小点距离交并比来计算边界框回归损失,更好地对齐预测框和真实目标框,提高了边界框回归的准确性;最后,通过在SPPCSPC层后添加感受野增强模块,增强了模型对小尺度害虫的识别能力。实验结果表明,改进后的YOLOv7模型平均准确度为80.4%,精确率为85.3%,召回率为75.1%,较改进前分别提升了3.4%、3.2%、2.6%。该模型对农业害虫具有更好的识别效果和鲁棒性,为农业害虫监测与防治提供了更准确和可靠的工具。