Automatic grape leaf diseases identification via UnitedModel based on multiple convolutional neural networks

Grape diseases are main factors causing serious grapes reduction.So it is urgent to develop an automatic identification method for grape leaf diseases.Deep learning techniques have recently achieved impressive successes in various computer vision problems,which inspires us to apply them to grape diseases identification task.In this paper,a united convolutional neural networks(CNNs)architecture based on an integrated method is proposed.The proposed CNNs architecture,i.e.,UnitedModel is designed to distinguish leaves with common grape diseases i.e.,black rot,esca and isariopsis leaf spot from healthy leaves.The combination of multiple CNNs enables the proposed UnitedModel to extract complementary discriminative features.Thus the representative ability of United-Model has been enhanced.The UnitedModel has been evaluated on the hold-out PlantVillage dataset and has been compared with several state-of-the-art CNN models.The experimental results have shown that UnitedModel achieves the best performance on various evaluation metrics.The UnitedModel achieves an average validation accuracy of 99.17%and a test accuracy of 98.57%,which can serve as a decision support tool to help farmers identify grape diseases.

Recognition of grape leaf diseases using MobileNetV3 and deep transfer learning

Timely diagnosis and accurate identification of grape leaf diseases are decisive for controlling the spread of disease and ensuring the healthy development of the grape industry.The objective of this research was to propose a simple and efficient approach to improve grape leaf disease identification accuracy with limited computing resources and scale of training image dataset based on deep transfer learning and an improved MobileNetV3 model(GLD-DTL).A pre-training model was obtained by training MobileNetV3 using the ImageNet dataset to extract common features of the grape leaves.And the last convolution layer of the pre-training model was modified by adding a batch normalization function.A dropout layer followed by a fully connected layer was used to improve the generalization ability of the pre-training model and realize a weight matrix to quantify the scores of six diseases,according to which the Softmax method was added as the top layer of the modified networks to give probability distribution of six diseases.Finally,the grape leaf diseases dataset,which was constructed by processing the image with data augmentation and image annotation technologies,was input into the modified networks to retrain the networks to obtain the grape leaf diseases recognition(GLDR)model.Results showed that the proposed GLD-DTL approach had better performance than some recent approaches.The identification accuracy was as high as 99.84%while the model size was as small as 30 MB.

Effects of the severity and timing of basal leaf removal on the amino acids profiles of Sauvignon Blanc grapes and wines

The effects of the severity and timing of leaf removal(LR)on the amino acids of Sauvignon Blanc grapes and wines were studied during the 2017 growing season.High-performance liquid chromatography(HPLC)was used to analyze the amino acids profiles of grape berries and wines.The basal leaves were removed at three time points(40,56 and 72 days after flowering,named LR40,LR56 and LR72,respectively)at two severity levels(one at which the first,third,and fifth basal leaves of each shoot were removed(50%level);and another at which the first six basal leaves were removed(100%level)).The results showed that leaf removal had little impact on total soluble solids(°Brix),titratable acidity,pH or berry weight.The LR72-50%treated grapes had higher berry weight,titratable acidity and°Brix than those of the other treatments.The highest concentrations of total amino acids and of total amino acids except proline were detected in LR72-50%treated grapes(2 952.58 and 2 764.36 mg L^-1,respectively);the lowest were detected in LR72-100%treated grapes(2 172.82 and 2 038.71 mg L^-1,respectively).LR72-50%treatment significantly promoted the synthesis of aspartic acid,serine,arginine,alanine,aminobutyric acid and proline at both severity levels for grapes,the concentrations of all of these amino acids were increased relative to the control concentrations.The LR72-50%,LR40-100%and LR72-100%treated wines had higher total amino acids concentrations and higher concentrations of some individual amino acids,such as arginine,alanine and serine,than did the control wines.Of all the amino acids studied,glycine,tyrosine,cysteine,methionine and lysine were not significantly influenced by the timing or severity basal defoliation in grapes and wines.The present study reveals the effects of the timing and severity of leaf removal on the amino acids profiles of grapes and wines.

Effects of leaf removal and cluster thinning on berry quality of Vitis vinifera cultivars in the region of Weibei Dryland in China

Leaf removal and cluster thinning were carried out prior to veraison to evaluate the effects on berry quality of two Vitis vinifera cultivars（Cabernet Sauvignon and Ugni Blanc） in the Weibei Dryland of China in 2013 and 2014, and comprehensive analysis of the chemical and volatile composition in berries was performed. The results showed that content of reducing sugar in both varieties was not affected while total acid was generally decreased by leaf removal and cluster thinning. The pH of berry juice was correspondingly higher in most treatment groups. Meanwhile, promoting effects on accumulation of total phenols, tannin in both varieties and total anthocyanins in Cabernet Sauvignon were found. As for monomeric anthocyanins, percentage of malvidin and its derivatives was decreased by leaf removal and cluster thinning. Besides, cinnamylated anthocyanins decreased with the intensity of cluster thinning. The accumulation of non-anthocyanin phenolics was similarly affected in the two varieties. Notably, cluster thinning was more effective on enhancing the phenolics content than leaf removal. The combination of middle level of leaf removal and cluster thinning was the most favor to the accumulation of phenolic acids. Furthermore, cluster thinning could also significantly enhance the synthesis of flavanols and stilbenes. Lastly, content and variety of aroma compounds in both grape varieties were also significantly affected by the treatments. The results provided a theoretical basis for a combination of leaf removal and cluster thinning to improve quality of grapes and wines.

葡萄叶片对空气中邻苯二甲酸酯的吸收累积特征

【目的】为探索葡萄叶片及其内部组成物质对空气中PAEs的吸收累积特征,阐明葡萄叶片从空气中吸收累积DBP、DEHP、DIBP的能力及规律。【方法】试验采用盆栽葡萄覆盖玻璃罩的方式,在玻璃罩内放置盛有DBP、DEHP、DIBP混合物的甲醇溶液培养皿,让PAEs自然挥发被植株吸收累积,研究叶片及其内部组成物质脂肪、糖、蛋白质对PAEs的吸收累积特征。【结果】3个品种葡萄叶片对DBP、DEHP和DIBP的累积量随着PAEs处理浓度的增高而增大,且不同污染水平各单体含量及总含量差异显著(P<0.05)。葡萄叶片对DEHP的累积量最高,在高污染处理中其累积量占比在92.90%~93.86%,其次为DBP和DIBP。叶片脂肪、糖、蛋白质中均检出DBP、DEHP和DIBP,各污染处理3种化合物的含量与空白差异显著(P<0.05),三者在脂肪中的累积量最多,最高可达87.06 mg/kg,在糖中最高仅为5.818 mg/kg,在蛋白质中最高仅为3.793 mg/kg。通过对葡萄叶片累积PAEs能力与组成物质的相关性进行分析,葡萄叶片PAEs累积量与脂肪中PAEs含量呈显著正相关相关,与糖和蛋白无显著相关性。【结论】葡萄叶片可从空气中吸收累积PAEs,其累积PAEs能力与脂肪密切相关。

Response of 14C-Salicylic Acid to Heat Stress After Being Fed to Leaves of Grape Plants

An experiment was conducted to investigate the response of salicylic acid as a second messenger to the heat stress in grape plants. For this purpose, all leaves of grape (Vitis vinifera×V. labrussa L. cv. Jingxiu) plants were removed except the 3rd, 4th, 5th, 6th, and 7th ones. The 5th leaf was fed with C-SA, and the 4th and 6th leaves were exposed to high 14 temperature at 40±0.5°C. It was observed that more C-SA transported out from the 5th leaf and the distribution of C-SA 14 14 in each organ of plant altered in response to heat stress. The accumulation of C-SA in both the 4th and 6th leaves being 14 exposed to high temperature was at least three times higher than that in control. The distribution of C-SA in other distal 14 leaves (the 3rd and 7th leaf) decreased, but more C-SA accumulated in stems adjacent to the 4th or 6th leaf exposed to 14 high temperature. In addition, there was more C-SA being transported upwards or downwards while the 4th and 6th 14 leaves were exposed to high temperature respectively. Therefore, our results suggested that SA was closely involved in signal transduction of heat stress in grape plants. However, the ratio of C radioactivity assayed after SA being extracted 14 to that of direct assay with apparatus was more than 70%, which indicated about 30% C was lost or catabolized during 14 transportation.

Effect of Gamma Irradiation Doses on Morphological and Biochemical Attributes of Grape Saplings

The demand of grape in Bangladesh is fulfilled through import from foreign countries. The fruits of local cultivars of grapes are sour and seeded. Development of seedless grape varieties having increased sweetness, higher yield with better nutritional quality is necessary to reduce the import dependency. The present research activities are the part of a grape improvement project. A pot experiment was conducted at the Bangladesh Institute of Nuclear Agriculture (BINA), Mymensingh, during June to November 2011 to determine the suitable gamma irradiation doses on growth, leaf area and biochemical characters of grape saplings. Three vegetative bud stages viz. bud initiation stage, 4-leaf stage and 8-leaf stage, and four doses of gamma irradiation viz. 0, 5, 10, and 15 Gy were used as treatments. The experiment was laid out in a Randomized Complete Block Design with four replications. Different irradiation doses and vegetative bud stages showed significant variations in respect of plant growth characters, leaf area, soluble protein and total sugar content. Interaction effects also had significant variations on most of the parameters studied. Higher doses of gamma irradiation had showed detrimental effect on grape saplings. Generally, increased in irradiation doses showed decreased and detrimental effects on most of the parameters under study. Maximum numbers and length of roots, total dry matter, leaf area and chlorophyll-a and chlorophyll-b content were found at 5 Gy irradiation dose. Total soluble protein and sugar content of leaf were found maximum at no irradiation and 15 Gy, respectively. Higher number of roots and length, total dry matter, leaf area, chlorophyll-a, and b and soluble protein content of leaf were observed at bud initiation stage while 8-leaf stage showed maximum total sugar of leaf. In the combined effect of gamma irradiation and vegetative bud stages, all parameters showed best results in 5 Gy with bud initiation stage except total sugar content of leaf.

阳光玫瑰葡萄需水量及耗水规律研究

为了明确阳光玫瑰葡萄生长期的需水量,为葡萄园节水灌溉提供理论依据,以阳光玫瑰葡萄为研究对象,通过调查生产上有效降雨量,计算水分盈亏量,在水分盈亏量的基础上设计3个灌水量处理(1 500、2 250、3 000 m^(3)/hm^(2)),研究不同灌水量对葡萄叶水势、树冠覆盖率、叶面积系数、果实品质和耗水特性的影响,确定葡萄需水量,并揭示其耗水规律。结果表明:在2 250、3 000 m^(3)/hm^(2)灌水量处理下,果实可溶性糖含量、可溶性固形物含量、固酸比均无显著性差异,但均显著高于1 500 m^(3)/hm^(2)灌水量处理;在2 250 m^(3)/hm^(2)灌水量下既不影响果实品质,又能节水;不同灌水量下葡萄耗水规律为果实膨大期>果实成熟期>新梢生长期>花期。综上,阳光玫瑰葡萄生长期总需水量为5 539 m^(3)/hm^(2)。

基于改进EfficientNetB0模型的葡萄叶部病害识别方法

为了高效、准确地识别葡萄叶部病害,文中提出了LE-EfficientNet模型,在EfficientNetB0模型基础上,采用大核注意力(LKA)机制替换原模型部分MBConv模块中的压缩激励网络(SENet),接着利用跳跃连接在最后一层卷积层后面融入高效通道注意力机制(ECA),结合三种注意力机制让网络更高效地提取葡萄叶部病害的局部重要信息,并引用Adam优化器替换原模型的SGD优化器,提升了分类模型的泛化能力。在PlantVillage葡萄叶部病害数据集上训练,结果表明,LE-EfficientNet模型相比原模型准确率提升了1.58%,总体精度提升了1.62%,召回率提升了1.46%,F_(1)分数提升了1.53%,并且参数量仅有10.18 MB,比原模型参数量降低2.7 MB,与其他经典网络模型相比,性能评估指标均有不同程度的提升,该研究为葡萄叶部病害识别提供了新的参考与借鉴。

基于改进MobileViT的葡萄叶部病害识别模型

本研究提出了一种优化的葡萄叶部病害识别模型CD-MobileViT。首先,将MobileViT作为基础网络,在Layer1、Layer2后面均嵌入坐标注意力模块CA(Coordinate Attention),以使网络能更有效地捕捉不同位置的关键特征;其次,在网络全连接层之后添加Dropout层,防止数据出现过拟合现象;最后,选用结合权重衰减的优化器AdamW(Adam with Weight Decay Regularization),更好地控制模型复杂度并提高泛化能力。实验结果显示,相较于MobileViT基础网络,改进后的CD-MobileViT网络在精确率、召回率、F1得分和准确率方面分别提高了1.77、1.85、1.65、1.75个百分点,与其他几种经典网络模型(InceptionV1、MobileNetV2、EfficientNetB0、VGG-16)相比也有不同程度的提升(0.25~1.47个百分点),说明本研究提出的模型在葡萄叶部病害识别上有良好的效果,未来可部署到移动端使用,为葡萄叶部病害的准确识别提供新的解决方案。

基于EBP-YOLOv8的葡萄叶病害检测与识别方法研究

为提高实际环境中葡萄叶病害检测的准确率,适合视频实时监测、无人机等嵌入式AI应用场景,对YOLOv8目标检测模型从模型结构、轻量化等方面进行改进,构建了EBP-YOLOv8。首先在颈部网络中引入BiFPN结构,加强模型特征层之间的融合,改善对小目标的检测能力;其次使用C2_P来替换颈部网络中的C2f结构,实现模型的轻量化,在降低模型计算量的同时而不影响其精度;然后在特征提取网络中融入EMA注意力机制,提升网络对感兴趣区域的关注,提升模型对复杂背景、相似病斑的识别能力;最后将CIoU损失函数替换为ECIoU损失函数,进一步提升模型的检测性能,使模型能够更好地收敛。EBP-YOLOv8对比YOLOv8n、Faster-RCNN、RetinaNet、YOLOv8n、YOLOv8s、YOLOv7、YOLOv7-Tiny、YOLOv4-Tiny,mAP分别提升了3.2%、13.87%、3.49%、3.2%、1.3%、5%、4.7%、8.8%,模型大小仅5.3MB。改进后的算法在轻量化及保证实时性的同时有效提高了检测精度,可以为开发葡萄叶病害实时检测边缘系统提供有效参考。

‘巨玫瑰’葡萄叶片原生质体的分离

以‘巨玫瑰’葡萄茎段为试材进行组培苗培养,通过调节IBA浓度,探究其生根诱导的最适培养基。并在最适培养基条件下,以无菌苗叶片为材料,对原生质体分离条件进行研究,为细胞融合、基因工程及品种改良等研究奠定基础。结果表明,IBA浓度对茎段生根有显著影响,以0.4 mg·L^(-1)时生根效果最好,根长为2.91 cm,增长系数为0.86;酶种类和浓度、酶解时间对原生质体分离有显著影响,采用40 g·L^(-1)纤维素酶+15 g·L^(-1)离析酶+5 g·L^(-1)果胶酶的酶液组合、酶解时间为2 h时,原生质体的产量和活性均达到最佳,分别为每克3.20×10^(6)个和82.88%。

圆叶葡萄和欧亚种葡萄叶片提取物对葡萄霜霉病的防治作用

为了明确不同抗性葡萄的叶片提取物对葡萄霜霉病的防治作用,以对葡萄霜霉病免疫的圆叶葡萄品种Noble及感病的欧亚种葡萄品种赤霞珠的幼叶和成龄叶乙醇提取物(叶片经干燥、粉碎后,以95%乙醇进行超声提取)为试材,设置不同浓度梯度,检测其对葡萄霜霉病菌(Plasmopara viticola,葡萄生单轴霉)孢子囊和游动孢子萌发的影响,同时测定其对霜霉病的室内防治作用。结果表明,两种抗性葡萄的叶提取物对葡萄霜霉病菌孢子囊和游动孢子的萌发均具有抑制作用。1 mg/mL叶片提取物处理下孢子囊萌发率即显著低于对照(无菌水和5%乙醇),而且叶片提取物浓度从10 mg/mL增加到50 mg/mL时孢子囊萌发率显著下降。在10 mg/mL叶片提取物处理下游动孢子萌发率即显著低于对照。两种抗性葡萄幼叶和成龄叶提取物浓度为10、25和50 mg/mL时,仅10 mg/mL“赤霞珠”成龄叶提取物对葡萄霜霉病的室内防治效果为95.0%,其余处理均为100.0%。“Noble”和“赤霞珠”叶提取物具有抑制霜霉病菌孢子囊和游动孢子萌发的作用,其室内防治效果显著。

施硒和6-BA对葡萄叶片衰老与活性氧代谢的影响

【目的】明确氨基酸硒和6-BA对葡萄叶片衰老与活性氧代谢的影响,为延缓叶片衰老技术的提出提供理论依据。【方法】以设施延迟栽培条件下叶片衰老速度不同的‘意大利’和‘无核白鸡心’2个葡萄品种为试材,分别进行叶面喷施氨基酸硒和6-BA处理,以喷施等量清水作为对照,研究施硒及6-BA对叶片衰老期间各功能叶片的叶绿素含量、净光合速率(Pn)、丙二醛(MDA)含量和超氧阴离子自由基(O2·)、过氧化氢(H2O2)含量及过氧化氢酶(CAT)、过氧化物酶(POD)、超氧化物歧化酶(SOD)抗氧化酶活性的影响。【结果】与对照相比,外源氨基酸硒和6-BA处理显著延缓了叶片的叶绿素含量和净光合速率(Pn)的下降,提高了过氧化氢酶(CAT)、过氧化物酶(POD)、超氧化物歧化酶(SOD)的活性,从而减慢了丙二醛(MDA)、超氧阴离子自由基(O2·)、过氧化氢(H2O2)的上升速率,延缓了叶片衰老。2个品种间比较,‘意大利’叶片衰老缓慢,在生育后期仍能维持较高的抗氧化酶活性。【结论】氨基酸硒和6-BA处理可有效提高叶片的抗衰老能力,延长叶片的功能期。

一种新的葡萄叶面积测定方法

为验证照相机辅助计算机图像处理技术在测定叶面积上的可行性,对少毛变叶葡萄、红提和藤稔的试管苗及大田植株叶面积进行测定,同时与网格法、复印称重法和直接称重法进行比较分析。结果表明:照相机辅助计算机图像处理技术测定法和上述几种测定方法测定结果无显著性差异,适用于少毛变叶葡萄、红提和藤稔试管苗及大田植株葡萄叶面积的测量,并具有准确、快速的优点。

设施葡萄常用品种的需冷量、需热量及2者关系研究

用3种不同的需冷量估算模型(≤7.2℃模型、0～7.2℃模型和犹他模型)和2种不同的需热量估算模型(生长度小时模型和有效积温模型)分别对22种设施葡萄常用品种的需冷量和需热量进行测定,同时分析2者的相互关系。结果表明,葡萄解除休眠的需冷量和萌芽展叶的需热量因品种和种不同而异。需冷量值品种间差异较大,介于573～1 246 h(≤7.2℃模型)或573～971 h(0～7.2℃模型)或917～1 090 C.U(犹他模型),且欧美杂种品种需冷量值普遍高于欧亚种品种;而需热量值品种间差异较小,介于9 976～12 541 GDH℃(生长度小时模型)或253～353 D℃(有效积温模型),且欧美杂种品种略低于欧亚种品种。同时研究表明葡萄的需冷量和需热量与其果实成熟期没有必然联系。无论以何种估算模型估算葡萄的需冷量和需热量,我国设施葡萄常用品种的需冷量和需热量之间均呈负相关关系。

6-BA及氨基酸硒对葡萄叶片衰老的影响

【目的】明确6-BA及氨基酸硒在激素水平上对葡萄叶片衰老的调控,为设施葡萄叶片衰老延缓技术的建立提供理论依据。【方法】在设施葡萄延迟栽培条件下,以叶片衰老速度不同的‘意大利’和‘无核白鸡心’2个葡萄品种为试材,分别进行叶面喷施6-BA和氨基酸硒处理,以清水为对照,测定不同处理和对照叶片衰老期间功能叶片的叶绿素含量和净光合速率(Pn)及内源激素含量与比值的变化。【结果】外源6-BA和氨基酸硒处理显著延缓了叶片叶绿素含量和净光合速率的下降,明显提高了玉米素核苷(ZR)和赤霉素(GA3)含量和ZR/ABA(脱落酸)、GA3/ABA、(ZR+GA3)/ABA比值,显著降低了ABA含量。生长素(IAA)具有前期保持叶片生长发育和后期促进衰老的双重作用。2个葡萄品种间比较,‘意大利’叶片衰老缓慢。【结论】6-BA和氨基酸硒通过维持较高的GA3/ABA、ZR/ABA和(GA3+ZR)/ABA比值,提高了葡萄叶片的叶绿素含量和净光合速率,延长了功能期,因此,外源喷施6-BA和氨基酸硒是延缓葡萄叶片衰老的重要技术措施。

H_2O_2对葡萄离体叶片白藜芦醇的诱导作用

【目的】葡萄叶片白藜芦醇含量较虎杖根含量低,难以作为药用提取用途,研究采用诱导剂升高葡萄离体叶片中白藜芦醇的含量。【方法】实验采用H_2O_2对离体红地球葡萄叶片进行诱导,采用薄层及高效液相色谱法分析白藜芦醇的含量变化,并使用RT-PCR方法分析芪合酶基因(STS)的表达量变化。【结果】H_2O_2对离体叶片芪类物质具有显著的诱导作用,诱导白藜芦醇的积累呈现显著的量时依赖性;5%H_2O_2处理48 h后叶片白藜芦醇含量可提高15倍以上,鲜重达到263μg/g。【结论】H_2O_2可以作为葡萄离体叶片白藜芦醇的有效诱导剂。

不同施肥深度对酿酒葡萄叶片养分和产量及品质的影响

以贺兰山东麓典型区御马酿酒葡萄基地6年生赤霞珠为供试材料,研究了酿酒葡萄叶片、产量及品质对3种施肥深度(20、40、60cm)下N、P、K配方施肥方式的响应。结果表明,酿酒葡萄叶片中氮含量远高于叶柄,叶片中氮素含量随着生育期的延长逐渐降低,而叶柄中氮含量随施肥深度增加和生育期延长而产生的变化很小,40cm的施肥深度能显著增加叶片中氮素含量;叶柄磷素平均含量高于叶片,全生育期叶片磷含量变化很小,而施肥60cm深度显著促进了叶片和叶柄中磷含量的提高;叶柄中钾含量远高于叶片,施肥深度对叶柄和叶片中钾含量未产生明显影响,但生育后期追施钾肥显著提高叶片钾含量;施肥深度达到40cm时酿酒葡萄的产量最高,而60cm的施肥深度产量显著下降。化肥适度深施到40cm有改善酿酒葡萄糖分的积累且增加糖酸比的趋势;而60cm的施肥深度总糖和糖酸比显著下降。由此说明,适度深施化肥到40cm左右的深度有利于提高肥料的利用率,从而促进葡萄的营养生长,提高产量且改善品质。

高温胁迫对葡萄叶片三项生理指标的影响

将葡萄幼苗置于高温胁迫条件下,通过测定叶片中脯氨酸、叶绿素和可溶性糖的含量来研究葡萄幼苗的抗热、抗水分胁迫的能力。结果表明:热激3h可溶性糖的含量达到最大值。葡萄叶片脯氨酸的含量在高温胁迫下明显高于对照,而叶绿素的含量和可溶性糖的含量则随胁迫时间的延长呈现“低—高—低”的波动。实验结果表明在短时间内葡萄幼苗具有快速抗热的机制。