Development of a Localized Dataset for Plant Disease Detection in Ohodo, Enugu State

In recent times in Nigeria, there has been incessant rise in food insecurity which has not only affected the individuals but has also drastically affected the stability of the nations economy. Plant diseases have been identified as one of the major challenges facing farmers worldwide leading to substantial crop losses and economic hardship. This research addresses the critical challenge of plant disease detection in Ohodo, Enugu state, Nigeria's agricultural sector by developing and curating a localized, high-quality dataset from Ohodo, Enugu State. Unlike previous studies that rely on generic, globally sourced datasets, this work emphasizes the need for an indigenous dataset that accurately represents the unique environmental conditions, crop varieties, and disease strains of a specific region. The core output of this foundational phase is a robust dataset consisting of 356 healthy and 366 unhealthy plant images, which provides a balanced and sufficient resource for training a machine learning model. The study proposes a Convolutional Neural Network (CNN) architecture designed to leverage this localized data. The methodology outlines a systematic approach to data collection, including stratified sampling, high-resolution image capture, and meticulous, expert-driven labeling to ensure data integrity. Ethical considerations, such as informed consent and community engagement, were central to the process, ensuring the research directly serves the needs of the local farming community. The primary finding confirms the successful establishment of this unique dataset, which serves as a critical first step towards creating a more accurate, generalizable, and practical plant disease identification system tailored to local agricultural realities, ultimately aiming to improve crop yields and farmer livelihoods in Ohodo.