randellmwania/Coffee-Image-Classification

Our project tackles the crucial task of detecting coffee leaf diseases like Rust, Cercospora, and Phoma, endangering Kenya's coffee farms. Using state-of-the-art technology and Arabica coffee leaf image data, our CNN models deliver impressive accuracy, providing farmers with early disease detection to protect their livelihoods. # **COFFEE LEAF DISEASE IMAGE CLASSIFICATION** Our project addresses the critical challenge of detecting coffee leaf diseases, including Rust, Cercospora, and Phoma, threatening Kenya's coffee farms. Leveraging cutting-edge technology and a dataset of Arabica coffee leaf images, our CNN models achieve remarkable accuracy, empowering farmers with early disease detection to safeguard their livelihoods. ### **Collaborators:** - Crystal Wanjiru - Randell Mwania - Victorine Imbuhila - Sadi Kiri - Ian Kedeyie - Simon Ng'ethe ## # **Business Understanding** Imagine the rolling hills of Kenya, vibrant with emerald coffee plantations. But a silent threat looms – Coffee rust, Cercospora, and Phoma, diseases capable of wiping out entire harvests and jeopardizing the livelihoods of thousands of small-scale farmers. This project rises to the challenge, wielding a powerful weapon: cutting-edge technology for early and accurate disease detection. By empowering farmers with this knowledge, we equip them to act swiftly. Timely application of targeted fungicides becomes their shield, minimizing crop losses and protecting their precious income. It's more than just protecting profits; it's safeguarding a cultural cornerstone, preserving the deep-rooted tradition of Kenyan coffee and ensuring the sustainability of the prized Arabica bean. This project is not just about numbers and yields; it's about people, communities, and the future of a cherished heritage. Join us in this fight to secure the verdant tapestry of Kenyan coffee farms, one healthy leaf at a time. # **Data Understanding** The dataset contains leaf images which were collected from Arabica coffee type and it shows three sets of Phoma, Rust and Cescospora images and one set of healthy images. The data was obtained from Dataset on Mendeley, Dataset on Mendeley and web scrapping with the 'Download All images' extension from the chrome web store. # **Data Preparation** The data pre-processing involved noise filterin …