Dataset of Citrus Canker Growth Rate

The hypothesis of the research was computer vision, image processing programs could be helpful in early detection and identification of citrus canker. For this purpose, initially the dataset was developed by inoculating healthy citrus leaves with disease causing organism (X. citri pv. citri) under natural field conditions. Inoculation was done on susceptible citrus cultivars, C. paradisi (Grapefruit). The leaves were inoculated by injecting the bacterial suspension into fully expanded, immature leaves with needleless syringes. The images were captured in Crop Diseases Research Institute (C.D.R.I.), National Agricultural Research Centre (NARC), Islamabad Pakistan regularly to measure the growth rate of citrus canker. The dataset is hosted by the Department of Computer Software Engineering, Military College of Signals-NUST Islamabad, Pakistan and acquired under the mutual cooperation of the NUST and C.D.R.I., NARC Pakistan. The inoculated leaves images were further categorized into different stages. It defines the temporal change in citrus canker growth rate. Canker symptoms developed were noted daily (from day 1 to day 21). From this dataset we can estimate and predict the disease prevalence and spread in citrus orchard over a specified time and can develop an alarming system for monitoring and control of the disease before it reaches its maximum loss. The dataset will be helpful for researchers for both plant pathologists and computer vision experts for classifying, detection and identification of citrus canker over specified time. The dataset was developed based on different growth stages thus it will be a novel way to monitor the disease spread and identify the threshold level of the disease and take precautionary measures before it completely damages the whole fruit plant. Additionally, the computer vision experts using implication of image processing, machine learning and deep learning techniques can design and build an early warning system by modeling the different disease stages and thus on site efficient robust online application can be generated which could be very useful both for farmers and agriculture department for warning and early detection system. The dataset has immense potential to be use in various field of application. Plant pathologist can use this data for measuring and identification of citrus canker and further explore the effect of other environmental variable on disease development.本研究的假设为：计算机视觉（Computer Vision）与图像处理（Image Processing）程序可辅助实现柑橘溃疡病的早期检测与识别。为此，研究团队首先在自然田间条件下，利用致病病原菌（X. citri pv. citri）接种健康柑橘叶片以构建该数据集。接种对象为易感柑橘品种柚（C. paradisi，葡萄柚），具体操作为采用无针注射器将细菌悬浮液注射至完全展开的未成熟叶片中。研究人员定期在巴基斯坦伊斯兰堡国家农业研究中心（National Agricultural Research Centre, NARC）下属的作物病害研究所（Crop Diseases Research Institute, C.D.R.I.）内拍摄图像，以监测柑橘溃疡病的病情发展速率。本数据集由巴基斯坦伊斯兰堡信号军事学院-国立科技大学（Military College of Signals-NUST Islamabad）计算机软件工程系托管，其获取过程得到了巴基斯坦国立科技大学与NARC下属C.D.R.I.的合作支持。随后将接种后的叶片图像划分为不同的病程阶段，以体现柑橘溃疡病病情发展速率的时序变化。研究人员每日（第1天至第21天）记录溃疡病症状的发展情况。借助该数据集，可估算并预测指定时间段内柑橘果园中的病害流行与传播态势，并开发出可在病害造成最大损失前开展监测与防控的预警系统。本数据集可帮助植物病理学家与计算机视觉领域研究者完成柑橘溃疡病的分类、检测与识别任务。本数据集基于不同病情生长阶段构建，为监测病害传播、确定病害阈值水平以及在全株果树遭受完全破坏前采取预防措施提供了全新的研究思路。此外，计算机视觉领域研究者可结合图像处理、机器学习与深度学习技术，通过对不同病害阶段进行建模，设计并构建早期预警系统，进而开发出高效可靠的现场在线应用程序，这对于农户与农业部门的病害预警与早期检测工作均具有重要实用价值。本数据集拥有广阔的应用前景：植物病理学家可利用该数据测量与识别柑橘溃疡病，并进一步探究其他环境变量对病害发展的影响。