美洲棘蓟马Echinothrips americanus Morgan,1913是新入侵我国的种类。本文报道了该虫各发育阶段的外部形态和分子鉴定结果。首先利用体视显微镜对美洲棘蓟马不同虫态的外部形态进行了观察和拍照;再选择线粒体COⅠ基因中一段约430bp的...美洲棘蓟马Echinothrips americanus Morgan,1913是新入侵我国的种类。本文报道了该虫各发育阶段的外部形态和分子鉴定结果。首先利用体视显微镜对美洲棘蓟马不同虫态的外部形态进行了观察和拍照;再选择线粒体COⅠ基因中一段约430bp的序列进行PCR扩增和测序,将所得序列用于蓟马种类的分子鉴定。使用不同的矩阵和系统发育构建方法对待鉴定的蓟马以及另外38种蓟马进行了聚类分析,结果表明:基于线粒体COⅠ基因第1和2编码位点,利用NJ法构建的系统发育树可以较好地区别不同蓟马种类,可作为调查我国蓟马种类和分布的快速方便方法。最后对美洲棘蓟马的基本生物学特性和防控对策进行了综述。展开更多
Tuta absoluta(Meyrick) originated in South America and is one of the most serious pests of tomatoes. It is also known to attack other solanaceous crops, including potato, eggplant, pepper, tobacco, and weedy species s...Tuta absoluta(Meyrick) originated in South America and is one of the most serious pests of tomatoes. It is also known to attack other solanaceous crops, including potato, eggplant, pepper, tobacco, and weedy species such as black nightshade. After accidental introduction into Spain in 2006, this pest spread rapidly throughout Afro-Eurasia and has become a major threat to tomato production worldwide. Here, we report the first record of T. absoluta as an invasive pest in China. It was found in tomato fields in Ili Kazakg Autonomous Prefecture, Xinjiang Uygur Autonomous Region(Ili, Xinjiang), China, and its occurrence was confirmed by both morphological and molecular approaches. In Ili, T. absoluta has been found to attack eggplant, potato, and black nightshade. We found the larvae generally mining and feeding on leaves and boring into tomato fruits, with multiple larvae sometimes observed in a single fruit. Its infestation levels differ among the tomato fields and host species. In all of the surveyed tomato fields, T. absoluta infested 100% of plants. In some of the fields, up to 90% of the eggplant and 100% of the potato plants were infested. Since no natural enemies were found under field conditions, suitable management practices are urgently needed to stop the further spread of this destructive pest in China.展开更多
针对当前大田环境条件下对害虫进行识别研究的不足,以南方蔬菜重大害虫为研究对象,探索了一种在大田环境下使用黄色诱捕板对蔬菜害虫进行监测计数的方法。在经典图像处理算法基础上,根据害虫监测目标的需要,提出了一种基于结构化随机森...针对当前大田环境条件下对害虫进行识别研究的不足,以南方蔬菜重大害虫为研究对象,探索了一种在大田环境下使用黄色诱捕板对蔬菜害虫进行监测计数的方法。在经典图像处理算法基础上,根据害虫监测目标的需要,提出了一种基于结构化随机森林的害虫图像分割算法和利用不规则结构的特征提取算法,进一步结合背景去除、干扰目标去除和检测模型计数子算法,集成设计了基于视觉感知的蔬菜害虫计数算法(Vegetable pest counting algorithm based on visual perception,VPCA-VP)。使用了现场环境下拍摄的图像进行实验与分析,共识别出蓟马9351只,烟粉虱202只,实蝇23只。经过与人工计数比对得出,本文基于视觉感知的蔬菜害虫计数算法的平均识别正确率为94.89%。其中,蔬菜害虫蓟马的识别正确率为93.19%,烟粉虱的识别正确率为91%,实蝇的识别正确率达到100%。算法达到了较好的测试性能,可以满足害虫快速计数需求,在农田害虫监测中有一定的应用前景。展开更多
【目的】智能虫情测报灯诱捕到的农业害虫因种类繁多、虫体姿态多样、鳞片脱落等原因造成有些害虫图像存在种间相似和种内差异的现象。为了提高农业灯诱害虫识别率,针对YOLOv4检测模型检测到且容易混淆的19种灯诱害虫,本文提出了基于双...【目的】智能虫情测报灯诱捕到的农业害虫因种类繁多、虫体姿态多样、鳞片脱落等原因造成有些害虫图像存在种间相似和种内差异的现象。为了提高农业灯诱害虫识别率,针对YOLOv4检测模型检测到且容易混淆的19种灯诱害虫,本文提出了基于双线性注意力网络的农业灯诱害虫细粒度图像识别模型。【方法】首先,根据灯诱害虫外观图像的相似性和检测误检的情况,将19种害虫分为6类;将所有害虫图像通过补边操作使得长宽相等,并缩放至统一尺寸224×224像素。为了提高模型的鲁棒性和泛化能力,对害虫图像进行镜像翻转、旋转180度、高斯噪声和均值滤波的数据增强,训练集、验证集和测试集样本量按照8﹕1﹕1比例划分。然后,针对6类19种农业灯诱害虫细粒度图像,建立了基于双线性注意力网络的农业灯诱害虫识别模型(bilinear-attention pest net,BAPest-net),模型包括双线性特征提取、注意力机制和分类识别3个模块;通过修改特征提取模块的下采样方式提高特征提取能力;添加注意力机制模块让整个模型更关注于局部细节的特征,将双线性结构中的上下两个注意力机制的输出进行外积运算增加细粒度特征的权重,提高识别的准确性和学习效率;模型优化器使用随机梯度下降法SGD,分类模块中使用全局平均池化,旨在对整个网络从结构上做正则化防止过拟合。最后,在同一个训练集训练VGG19、Densenet、ResNet50、BCNN和BAPest-net 5个模型,对6类相似的19种农业灯诱害虫进行识别,以精准率、Precision-Recall(PR)曲线和平均识别率作为模型的评价指标。【结果】BAPest-net对6类相似的19种农业灯诱害虫平均识别率最高,达到94.9%;BCNN次之,为90.2%;VGG19模型最低,为82.1%。BAPest-net识别的6类害虫中4类鳞翅目害虫的平均识别率均大于95%,表明该模型能较好地识别出鳞翅目害虫。测试结果中仍存在少�展开更多
文摘美洲棘蓟马Echinothrips americanus Morgan,1913是新入侵我国的种类。本文报道了该虫各发育阶段的外部形态和分子鉴定结果。首先利用体视显微镜对美洲棘蓟马不同虫态的外部形态进行了观察和拍照;再选择线粒体COⅠ基因中一段约430bp的序列进行PCR扩增和测序,将所得序列用于蓟马种类的分子鉴定。使用不同的矩阵和系统发育构建方法对待鉴定的蓟马以及另外38种蓟马进行了聚类分析,结果表明:基于线粒体COⅠ基因第1和2编码位点,利用NJ法构建的系统发育树可以较好地区别不同蓟马种类,可作为调查我国蓟马种类和分布的快速方便方法。最后对美洲棘蓟马的基本生物学特性和防控对策进行了综述。
基金This work was supported by the National Key Research and Development Program of China(2017YFC1200600,2016YFC1201200)the Science and Technology Innovation Program of Chinese Academy of Agricultural Sciences(caascx-2017-2022-1AS).
文摘Tuta absoluta(Meyrick) originated in South America and is one of the most serious pests of tomatoes. It is also known to attack other solanaceous crops, including potato, eggplant, pepper, tobacco, and weedy species such as black nightshade. After accidental introduction into Spain in 2006, this pest spread rapidly throughout Afro-Eurasia and has become a major threat to tomato production worldwide. Here, we report the first record of T. absoluta as an invasive pest in China. It was found in tomato fields in Ili Kazakg Autonomous Prefecture, Xinjiang Uygur Autonomous Region(Ili, Xinjiang), China, and its occurrence was confirmed by both morphological and molecular approaches. In Ili, T. absoluta has been found to attack eggplant, potato, and black nightshade. We found the larvae generally mining and feeding on leaves and boring into tomato fruits, with multiple larvae sometimes observed in a single fruit. Its infestation levels differ among the tomato fields and host species. In all of the surveyed tomato fields, T. absoluta infested 100% of plants. In some of the fields, up to 90% of the eggplant and 100% of the potato plants were infested. Since no natural enemies were found under field conditions, suitable management practices are urgently needed to stop the further spread of this destructive pest in China.
文摘针对当前大田环境条件下对害虫进行识别研究的不足,以南方蔬菜重大害虫为研究对象,探索了一种在大田环境下使用黄色诱捕板对蔬菜害虫进行监测计数的方法。在经典图像处理算法基础上,根据害虫监测目标的需要,提出了一种基于结构化随机森林的害虫图像分割算法和利用不规则结构的特征提取算法,进一步结合背景去除、干扰目标去除和检测模型计数子算法,集成设计了基于视觉感知的蔬菜害虫计数算法(Vegetable pest counting algorithm based on visual perception,VPCA-VP)。使用了现场环境下拍摄的图像进行实验与分析,共识别出蓟马9351只,烟粉虱202只,实蝇23只。经过与人工计数比对得出,本文基于视觉感知的蔬菜害虫计数算法的平均识别正确率为94.89%。其中,蔬菜害虫蓟马的识别正确率为93.19%,烟粉虱的识别正确率为91%,实蝇的识别正确率达到100%。算法达到了较好的测试性能,可以满足害虫快速计数需求,在农田害虫监测中有一定的应用前景。
文摘【目的】智能虫情测报灯诱捕到的农业害虫因种类繁多、虫体姿态多样、鳞片脱落等原因造成有些害虫图像存在种间相似和种内差异的现象。为了提高农业灯诱害虫识别率,针对YOLOv4检测模型检测到且容易混淆的19种灯诱害虫,本文提出了基于双线性注意力网络的农业灯诱害虫细粒度图像识别模型。【方法】首先,根据灯诱害虫外观图像的相似性和检测误检的情况,将19种害虫分为6类;将所有害虫图像通过补边操作使得长宽相等,并缩放至统一尺寸224×224像素。为了提高模型的鲁棒性和泛化能力,对害虫图像进行镜像翻转、旋转180度、高斯噪声和均值滤波的数据增强,训练集、验证集和测试集样本量按照8﹕1﹕1比例划分。然后,针对6类19种农业灯诱害虫细粒度图像,建立了基于双线性注意力网络的农业灯诱害虫识别模型(bilinear-attention pest net,BAPest-net),模型包括双线性特征提取、注意力机制和分类识别3个模块;通过修改特征提取模块的下采样方式提高特征提取能力;添加注意力机制模块让整个模型更关注于局部细节的特征,将双线性结构中的上下两个注意力机制的输出进行外积运算增加细粒度特征的权重,提高识别的准确性和学习效率;模型优化器使用随机梯度下降法SGD,分类模块中使用全局平均池化,旨在对整个网络从结构上做正则化防止过拟合。最后,在同一个训练集训练VGG19、Densenet、ResNet50、BCNN和BAPest-net 5个模型,对6类相似的19种农业灯诱害虫进行识别,以精准率、Precision-Recall(PR)曲线和平均识别率作为模型的评价指标。【结果】BAPest-net对6类相似的19种农业灯诱害虫平均识别率最高,达到94.9%;BCNN次之,为90.2%;VGG19模型最低,为82.1%。BAPest-net识别的6类害虫中4类鳞翅目害虫的平均识别率均大于95%,表明该模型能较好地识别出鳞翅目害虫。测试结果中仍存在少�
