ABSTRACT
Potatoes are a crucial crop worldwide, yet they face significant threats from various diseases that can harm yield and quality. Prompt identification and treatment of these diseases are vital for food security and agricultural productivity. Convolutional Neural Networks (CNNs), a form of deep learning, have demonstrated promising capabilities in detecting plant diseases via image-based analysis. This research aims to identify early disease indicators in potato leaves using CNN. The proposed system will accurately classify new images as healthy or diseased after being trained on a dataset comprising photos of healthy and diseased potato leaves. Early detection enables farmers to take swift action, curbing disease spread and minimizing crop losses. The suggested system could serve as a valuable tool for monitoring crop health and implementing necessary precautions to ensure a robust harvest. The study utilized the Potato Disease Leaf Dataset (PLD), encompassing 4072 raw images across training, testing, and validation sets, with three classes: Healthy, Early Blight, and Late Blight. The CNN algorithm achieved high accuracy, reaching 98.2% when trained on raw potato leaf images from different classes. Future endeavors may involve implementing this model through APIs and GUIs to enable early disease detection in potato leaves, providing significant benefits for farmers.
