UGV - Guava Leaves Disease Dataset Bangladesh

The Guava Leaf Image Dataset is a curated collection of annotated guava leaf images developed at the University of Global Village (UGV), Barishal, Bangladesh. It aims to support research in smart agriculture, particularly in plant disease detection using computer vision and machine learning. Guava (Psidium guajava), a widely cultivated tropical fruit, is vulnerable to various diseases that affect crop yield and quality. This dataset offers a structured image resource to facilitate the development of automated diagnostic tools.

The dataset is organized into seven folders: Canker, Curling, Rust, Healthy, Nutritional Deficiency, Leaf Spot, and Powdery Mildew. Each folder contains images of guava leaves classified based on visible symptoms. These categories reflect common diseases and conditions that affect guava plants, allowing for the training and evaluation of AI models.

Canker includes leaves showing raised, dark brown lesions caused by bacterial infection. Curling represents leaves deformed due to viral attacks or environmental stress. Rust features leaves with orange-yellow powdery fungal spores. Healthy contains leaves with no visible damage, serving as a baseline. Nutritional Deficiency includes yellowed or browning leaves caused by lack of essential nutrients. Leaf Spot captures leaves with dark spots due to fungal infections. Powdery Mildew contains leaves with white, dusty fungal growth.

Images were taken in natural daylight using digital and smartphone cameras, ensuring realistic field conditions. Expert validation by plant pathologists at UGV ensures accurate labeling. The dataset supports machine learning tasks like image classification, model training, and precision agriculture applications.

Though limited to one geographic region, the dataset is a valuable tool for researchers, students, and developers working on crop disease detection. Future versions could include metadata, multi-angle images, and samples from diverse regions. Overall, the dataset provides a practical and scientifically grounded resource for improving guava farming through technology.

番石榴叶片图像数据集（Guava Leaf Image Dataset）是由孟加拉国巴里萨尔的全球村大学（University of Global Village, UGV）精心构建的带标注番石榴叶片图像精选集，旨在为智慧农业领域的研究提供支撑，尤其是基于计算机视觉（Computer Vision）与机器学习（Machine Learning）的植物病害检测方向。番石榴（Psidium guajava）是广泛种植的热带水果，易受多种病害侵染，进而降低作物产量与品质。本数据集提供结构化的图像资源，助力自动化诊断工具的开发。
该数据集共分为7个文件夹，分别为溃疡病（Canker）、卷叶病（Curling）、锈病（Rust）、健康叶片（Healthy）、营养缺失症（Nutritional Deficiency）、叶斑病（Leaf Spot）与白粉病（Powdery Mildew）。每个文件夹内均包含依据可见症状分类的番石榴叶片图像，这些类别涵盖了影响番石榴植株的常见病害与异常状况，可用于AI模型的训练与评估。
溃疡病文件夹收录由细菌感染引发的、带有隆起深褐色病斑的叶片；卷叶病文件夹收录因病毒侵染或环境胁迫导致形态变形的叶片；锈病文件夹收录带有橙黄色粉状真菌孢子的叶片；健康叶片文件夹收录无可见损伤的叶片，作为基准对照组；营养缺失症文件夹收录因缺乏必需养分导致黄化或褐变的叶片；叶斑病文件夹收录由真菌感染引发暗斑的叶片；白粉病文件夹收录带有白色粉状真菌附着物的叶片。
所有图像均在自然日光下使用数码相机与智能手机拍摄，以还原真实的田间拍摄环境。全球村大学的植物病理学家完成了专业标注校验，确保标签的准确性。本数据集可支撑图像分类、模型训练以及精准农业应用等机器学习任务。
尽管该数据集仅覆盖单一地理区域，但仍是面向作物病害检测领域的研究人员、学生与开发者的宝贵工具。未来版本可考虑添加元数据、多视角图像以及来自不同地域的样本。总体而言，本数据集为通过技术手段提升番石榴种植水平提供了兼具实用性与科学严谨性的资源。