Potato Viral Disease Dataset on both Foliar and Tuber

Potato (Solanum tuberosum) is the main crop that is vegetatively propagated(cloning). It is the third most important food crop after wheat and rice in terms of human consumption. This crop is affected by various types of diseases like bacterial, fungal, and viral diseases. As, a traditional practice some viral disease are identified by visual symptoms and some by laboratory methods like Enzyme-linked immunosorbent assay (ELISA), real time reverse-transcription- polymerase chain reaction (RT-PCR) etc., the above methods are tedious, computational costs along with time will be high and it is requiring a more labor-intensive and controlled lab structure to carry out the experiment. So, there is an urgent need for developing an automated model which employs machine learning, deep learning methodology which are subset of Artificial Intelligence. This model helps agriculturist/ farmers for early detection of viral disease. and take effective measures for huge yield loss. The dataset shows different types of viral diseases that the potato crop foliar and tubers get effected with namley Mosaic Virus, Potato Leaf Roll Virus(PLRV), on foliar, and on tuber Potato Spindle Tuber Viriod (PSTVD), Potato Virus Y(PVY)-tuber cracking. The dataset of foliar and tuber comprises of Nineteen hundred and seventy two images Mosaic(666), PLRV(527), Healthy leaf(135), PSTVD(85), PVY cracking (559). Each image of size in pixels is 4288*2848 and camera employed is Nikon D90,its configuration details include f-stop-f/5.3, ISO speed -ISO-250, focal length-80mm, contrast -medium, No flash. The dataset is collected from potato fields of University of Agricultural Sciences, Dharwad where 3 acres of land is cultivated with certified seeds from modipuram regional Centre and 1 acre is cultivated with uncertified seeds. The scientist from the research institute shared the image data from Indian Council of Agricultural Research- Central Potato Research Institute (ICAR- CPRI) shimla. The data set is captured in the day light by placing the leaf or tuber on black background. The image data collected has different symptoms like plant affected with mosaic virus has yellow color spread throughout the leaf the leaf can be classified as mild, medium and severe mosaic. The color feature plays a major role in classifying the mosaic viral disease the symptoms include yellow color propagated throughout the leaf. The PLRV symptoms are rolling of leaf towards upward direction the texture of the leaf is little crunchy in nature. The main features here are shape and texture for accurately classifying the disease. In tuber PSTVD where the tubers are elongated that is based on shape features the classifier gives the accurate results. PVY cracking it is one of the strains of potato virus Y where the tubers have cracks on external part which is noninfectious. The cracks generally start at bud and extend lengthwise.

马铃薯（Solanum tuberosum）是主要的无性繁殖（克隆繁殖）作物。按人类食用量计算，它是仅次于小麦和水稻的第三大粮食作物。该作物会受到多种病害侵扰，包括细菌性、真菌性及病毒性病害。传统上，部分病毒性病害通过视觉症状识别，部分则通过酶联免疫吸附试验（ELISA）、实时逆转录聚合酶链反应（RT-PCR）等实验室方法检测，但上述方法流程繁琐，计算成本与时间投入均较高，且需要耗费大量人力并搭建受控实验室环境开展实验。因此，亟需开发一种基于人工智能子集的机器学习、深度学习方法的自动化检测模型，助力农业从业者/农户实现病毒性病害的早期检测，从而采取有效措施避免大幅减产损失。本数据集涵盖马铃薯叶部与块茎所感染的多种病毒性病害，具体包括叶部感染的花叶病毒（Mosaic Virus）、马铃薯卷叶病毒（PLRV），以及块茎感染的马铃薯纺锤块茎类病毒（PSTVD）、马铃薯Y病毒（PVY）-块茎开裂型。其中叶部与块茎数据集共包含1972张图像：花叶病毒感染样本666张、PLRV感染样本527张、健康叶片样本135张、PSTVD感染样本85张、PVY块茎开裂型样本559张。所有图像的像素尺寸均为4288×2848，拍摄设备为尼康D90（Nikon D90），拍摄参数包括：光圈f/5.3、ISO感光度250、焦距80mm、对比度中等、未使用闪光灯。本数据集采集自印度达瓦德农业科学大学的马铃薯田，其中3英亩土地种植的是来自莫迪普拉姆区域中心的认证种子，1英亩种植的为非认证种子。研究机构的科研人员分享了来自印度农业研究委员会-中央马铃薯研究所（ICAR-CPRI）西姆拉分院的图像数据。所有图像均在日光下拍摄，拍摄时将叶片或块茎置于黑色背景之上。本次采集的图像包含多种病害症状：感染花叶病毒的植株叶片会出现全域黄色扩散，可分为轻度、中度、重度花叶三类；颜色特征在花叶病毒病害的分类中发挥关键作用，其典型症状为叶片全域出现黄色蔓延。PLRV感染的叶片会向上卷曲，叶片质地略带脆硬感，此时形状与纹理特征是准确分类该病害的核心要素。块茎感染PSTVD后会呈现细长形态，基于形状特征可实现精准分类。PVY块茎开裂型是马铃薯Y病毒的一个毒株，染病块茎的外部会出现裂纹，该症状不具备传染性，裂纹通常始于芽眼并沿纵向延伸。