New tools for studying epidemiology and resistance of grape powdery mildew to DMI fungicides

Using a Random Amplified Polymorphic DNA (RAPD) assay, we investigated the genetic polymorphism existing among 62 European isolates of the grape powdery mildew fungus (Uncinula necator [Schw.] Burr.). Isolates overwintering as mycelium in buds were genetically distinct from isolates overwintering as ascospores, suggesting the existence of two genetically isolated powdery mildew populations, and consequently of two independent sources of inoculum in the vineyard. Isolates resistant to fungicides inhibiting sterol 14α-demethylation (DMIs) were found in both populations, suggesting that resistance to DMIs may arise independently in the two powdery mildew populations. A PCR assay targeting the gene encoding U. necator 14α-demethylase has been developed which will permit an early, specific detection of U. necator infections, and may be useful for spraying programmes. ©1997 SCI     

A real‐time PCR for the detection of hepatopancreatic parvovirus (HPV) of penaeid shrimp

Hepatopancreatic parvovirus (HPV) causes a common shrimp disease that occurs in many shrimp farming regions, especially in the Indo Pacific, and infects most of the cultured penaeid species. There are seven geographic HPV isolates known, so a method to detect different HPV types is needed. We developed a sensitive and generic real‐time PCR assay for the detection of HPV. A pair of primers and TaqMan probe based on an HPV sequence obtained from samples of Fenneropenaeus chinensis from Korea were selected, and they were used to amplify a 92 bp DNA fragment. This real‐time PCR was found to be specific to HPV and did not react with other shrimp viruses. A plasmid (pHPV‐2) containing the target HPV sequence was constructed and used for determination of the sensitivity of this assay. The assay could detect a single copy of plasmid DNA, and it was used successfully in finding HPV in shrimp samples from the China‐Yellow Sea region, Taiwan, Korea, Thailand, Madagascar, New Caledonia and Tanzania.     

基于改进YOLOv7的油茶果实成熟度检测

为确保油茶果实处于最佳成熟度进行采摘,提高油茶果实的出油率及茶油品质,该研究针对自然环境下油茶果实多被遮挡的问题,以原始YOLOv7模型为基础进行改进,提出一种油茶果实成熟度检测方法。首先,在主干网络中引入十字交叉注意力机制（criss-cross attention,CCA）加强对被枝叶遮挡果实成熟度特征的提取能力;其次,使用基于距离和交并比的非极大值抑制（distance-iou non-maximum suppression,DIoU-NMS）算法代替传统非极大值抑制（nonmaximum suppression,NMS）算法,从而加强模型对相互遮挡果实的检测能力;最后,以训练集中3 098张油茶果实图像训练改进的YOLOv7模型,验证集中442张图像用于在训练过程中评估模型,并对测试集中885张图像进行测试。改进后的YOLOv7模型在测试集下的精确率P为93.52%,召回率R为90.25%,F1分数为91.86%,平均精度均值mAP为94.60%,平均检测时间为0.77 s,模型权重大小为82.6 M。与Faster R-CNN、EfficientDet、YOLOv3、YOLO...     

基于轻量化YOLOv4的生猪目标检测算法

针对目前生猪目标检测算法模型较大,实时性差导致其难以在移动终端中应用等问题,将一种改进的轻量化YOLOv4算法用于生猪目标检测.在群养猪环境下以不同视角和不同遮挡程度拍摄生猪图像,建立生猪目标检测数据集.基于轻量化思想,在YOLOv4基础上缩减模型大小.结果表明,本研究算法的准确率和召回率分别为96.85％和91.75...     

牛妊娠相关糖蛋白6快速检测试纸条的制备及应用

奶牛妊娠的早期诊断至关重要,妊娠相关糖蛋白6(PAG6)是母体胎盘组织分泌至外周循环中的特异蛋白质,因此,本研究致力于制备牛妊娠相关糖蛋白6快速检测试纸条。制备牛妊娠相关糖蛋白6多克隆抗体,将其稀释后加样至NC膜上,通过固定和封闭制备出一种可检测boPAG6的快速检测试纸条,在试纸条上加入待检血清进行孵育后洗涤,再加入二抗孵育后洗涤,最后加入显色液显色后进行快速检测。标准化快速检测试纸条是将抗体用PBS稀释成10μg/mL点样至NC膜上,在37℃温箱中固定抗体30 min,经5%的脱脂奶粉封闭后制成。待检血清滴加至试纸条后,在37℃温箱中孵育1 h后,进行洗涤和二抗孵育,最后在DAB显色试剂盒作用下显色3 min即可判定。临床检测的初步应用表明,在未孕与怀孕奶牛的混合血清中仍然可以判定是否怀孕,快速检测试纸条特异性强;血清在稀释3200倍的最低检测限时,快速检测试纸条的敏感性良好;而且该结果与ELISA的符合率在80%之上。试验重复性良好,并能在4℃和-20℃条件下保存2周。建立了一种奶牛早期妊娠诊断的快速检测试纸条并应用于初步的妊娠诊断。     

基于改进Harris角点检测的虚拟樱桃叶片重建方法

为了分析樱桃树叶片的形态结构,为樱桃树冠层光照分布及樱桃果树整形修剪提供理论基础,提出了一种基于改进Harris角点检测及NURBS曲线的樱桃树叶片重建方法。利用中值滤波法及经典边缘检测算法对获得的原始樱桃树叶片图像进行预处理,得到树叶轮廓。针对传统NUBRS曲线原理重构轮廓时无法构成闭合曲面,特征点之间相互干扰等问题,提出了一种新的算法来重构轮廓。该算法首先将特征点分为左右两部分,分别重建左右两侧轮廓,再将两者连接得到完整的轮廓;其次运用改进的Harris角点检测法提取角点来作为特征点;再次,通过检测窗口中心点灰度与其周围n邻域内其他像素点灰度的相似程度,计算灰度之差来设定一个阈值,并根据该阈值范围提取角点;最后,根据虚拟轮廓构建虚拟叶片。实验结果表明,改进后的算法大大减少了计算量,平均消耗时间由4.61s减少到2.30s。本文方法较真实地重构了樱桃树叶片边缘形状,为樱桃树冠层光照分布计算提供了技术支持。     

基于可见-近红外光谱的鲜食葡萄成熟品质关键指标检测

酚类物质是评价葡萄成熟品质的重要指标,本文利用可见-近红外光谱技术结合化学计量学定量分析方法对葡萄皮总酚、籽总酚、皮单宁和籽单宁含量开展了无损检测研究。通过手持式可见-近红外光谱仪采集巨玫瑰葡萄波长400～1 029 nm范围内的漫反射光谱,采用SPXY算法将其划分为校正集和预测集,结合标准正态变换(Standard normal variate, SNV)、多元散射校正(Multiplicative scatter correction, MSC)、一阶导数(First derivative, 1D)、二阶导数(Second derivative, 2D)、Savitzky-Golay卷积平滑(Savitzky-Golay smoothing,SG)和Savitzky-Golay卷积平滑+一阶导数(SG+1D)6种预处理方法以及偏最小二乘回归(Partial least squares regression, PLSR)、支持向量机回归(Support vector machine regression, SVR)和卷积神经网络(Convolutional neural networ...     

基于改进YOLO v5的复杂环境下桑树枝干识别定位方法

为实现复杂自然环境下对桑树嫩叶处枝干的识别检测,改变当前桑叶采摘设备作业过程中依赖人工辅助定位的现状,解决识别目标姿态多样和环境复杂导致的低识别率问题,提出一种基于改进YOLO v5模型的桑树枝干识别模型(YOLO v5-mulberry),并结合深度相机构建定位系统。首先,在YOLO v5的骨干网络中加入CBAM(Convolutional block attention module)注意力机制,提高神经网络对桑树枝干的关注度;并增加小目标层使模型可检测4像素×4像素的目标,提高了模型检测小目标的性能;同时使用GIoU损失函数替换原始网络中的IoU损失函数,有效防止了预测框和真实框尺寸较小时无法正确反映预测框及真实框之间位置关系的情况;随后,完成深度图和彩色图的像素对齐,通过坐标系转换获取桑树枝干三维坐标。试验结果表明：YOLO v5-mulberry检测模型的平均精度均值为94.2%,较原模型提高16.9个百分点,置信度也提高12.1%;模型室外检测时应检测目标数53,实际检测目标数为48,检测率为90.57%;桑树嫩叶处枝干三维坐标识别定位系统的定位误差为(9.498 5 mm...