Deep learning pre-trained model for maize leaves disease detection and diagnosis in tropical agriculture

Abstract

Traditional disease detection and diagnostic methods are extremely slow, inaccurate, and laborious. Recent advancements in technology have shifted focus towards employing sophisticated tools like deep learning to improve the effectiveness and precision of disease management in crops. The setback of plant leaf diseases on contemporary agriculture is a pressing issue. Maize, a fundamental crop, is essential for global food security, yet it is increasingly threatened by a range of pets that redefine plant health. The study explores deep learning techniques for detection and diagnosis of diseases in maize leaves. By utilizing Convolutional Neural Networks (CNNs), the study analyzes high-resolution maize leaf images to facilitate automated disease identification. The deep learning model is trained on both local and public datasets for the detection of diseases such as Blight, Common Blight and Gray Leaf Spot. A prediction accuracy of 96% was attained after extensive training of the deep learning model. The findings highlight the deep learning model’s capability to accurately identify and diagnose maize leaf diseases, illustrating its potential as an invaluable resource for farmers and agronomists. Incorporating this technology into precision agriculture can lead to earlier disease detection, timely interventions, and improved crop yields, fostering sustainable agricultural practices.