Image Dataset on Chili Leaf Diseases in the Krishna River Basin of the Deccan Plateau, India

Overview
This dataset is a comprehensive resource for researchers and professionals in agriculture, machine learning, and computer vision, focusing on chili plant disease detection and growth stage classification. It provides high-resolution images of both healthy and diseased chili leaves, as well as samples representing various growth stages, making it highly suitable for deep learning and AI-based precision agriculture applications.
Data Collection
The images were systematically collected between October 2024 and February 2025 from chili plantations in the Bellary and Raichur districts of Karnataka, and the Prathipadu, Vatticherukuru, and Medikonduru of Guntur and Parchur, Yeddana Pudi, and MarturPrakasam districts of Andhra Pradesh, which are recognized as the global epicenters for red chili production. All field collection was conducted under the direct supervision of agricultural experts to ensure the pathological accuracy of the healthy and diseased leaf samples. The collection effort spanned across diverse agro-climatic zones of the Krishna River Basin within the Deccan Plateau.
This multi-regional coverage ensured that a wide range of environmental and cultivation conditions were represented, capturing real-world variability in chili plant health, disease manifestation, and growth patterns.
Using advanced digital cameras, high-resolution images were gathered, documenting different disease types and developmental stages to support robust AI-driven agricultural research.
Dataset Structure
The dataset is organized into two primary categories:

Chili Leaf Disease Dataset
This category includes images depicting six major leaf diseases affecting chili plants. Each image is meticulously labeled and verified by agricultural experts to enable precise classification and model training.

Original Dataset: 1,856 high-resolution images (.jpg)
Augmented Dataset: 12,000 enhanced images generated through data augmentation techniques (rotation, flipping, contrast adjustment, and zoom)
# Disease Classes Covered:
Bacterial Spot

Curl Virus

Cercospora Leaf Spot

Nutrition Deficiency

White Spot

Healthy Leaves
Specifically, this dataset enables:
✅ Early and Accurate Disease Detection — facilitating the identification of key foliar diseases such as Bacterial Spot, Curl Virus and Cercospora Leaf Spot at early stages, thereby preventing large-scale crop losses and improving disease management efficiency.
✅ Automated Growth Stage Monitoring — providing visual evidence of chili plants at various growth stages, enabling automated phenotyping, yield estimation and adaptive farm management strategies using AI and computer vision models.
✅ Benchmark for AI and Computer Vision Models — serving as a standardized dataset for training and benchmarking deep learning architectures such as CNNs, Vision Transformers and hybrid models for plant disease recognition, segmentation, and classification tasks.

概览
本数据集为农业、机器学习与计算机视觉领域的科研人员及行业从业者提供了一套综合性资源，核心聚焦于辣椒病害检测与生育期分类任务。数据集包含健康与染病辣椒叶片的高分辨率图像，以及覆盖不同生育期的样本，非常适用于深度学习与人工智能驱动的精准农业相关应用。
数据采集
图像采集工作于2024年10月至2025年2月间系统性开展，采集地点涵盖卡纳塔克邦贝拉里（Bellary）、莱丘尔（Raichur）县，以及安得拉邦贡图尔县的普拉蒂帕杜（Prathipadu）、瓦蒂切鲁库鲁（Vatticherukuru）、梅迪孔杜鲁（Medikonduru），帕丘尔（Parchur）、耶德纳普迪（Yeddana Pudi）与马尔图尔（Martur）所在的普拉卡萨姆县——上述区域均为全球公认的红辣椒主产区核心地带。所有田间采集工作均由农业专家全程监督，以确保健康与染病叶片样本的病理准确性。本次采集覆盖了德干高原内克里希纳河流域的多样农业气候区。
这种多区域覆盖范围确保了数据集涵盖了多样的环境与种植条件，能够捕捉辣椒植株健康状况、病害表现与生长模式的真实田间变异性。本次采集采用先进数码相机获取高分辨率图像，记录了不同病害类型与生育期的样本，可为人工智能驱动的农业研究提供坚实支撑。
数据集结构
本数据集主要分为两大类别：
辣椒叶片病害数据集
本类别包含侵染辣椒植株的六大主要叶部病害图像。所有图像均经农业专家精细标注与核验，可支撑精准分类与模型训练。
原始数据集：1856张高分辨率JPEG（.jpg）图像
增强数据集：通过旋转、翻转、对比度调整与缩放等数据增强技术生成的12000张增强图像
覆盖病害类别：
细菌性斑点病（Bacterial Spot）
曲叶病毒病（Curl Virus）
尾孢叶斑病（Cercospora Leaf Spot）
营养缺素症（Nutrition Deficiency）
白斑病（White Spot）
健康叶片（Healthy Leaves）
具体而言，本数据集可实现以下应用场景：
✅ 早期精准病害检测：可实现细菌性斑点病、曲叶病毒病与尾孢叶斑病等主要叶部病害的早期识别，从而避免大规模作物减产，提升病害防控效率。
✅ 自动化生育期监测：提供辣椒植株不同生育期的可视化样本，可借助人工智能与计算机视觉模型实现自动化表型分析、产量预估与自适应农田管理策略的构建。
✅ 人工智能与计算机视觉模型基准数据集：可作为标准化数据集，用于训练与基准测试卷积神经网络（Convolutional Neural Networks, CNNs）、视觉Transformer（Vision Transformers）及混合模型等深度学习架构，以完成植物病害识别、分割与分类任务。