Proposed Convolutional Neural Network for the Detection of Guava Leaf Diseases

Fruit cultivation is essential for domestic consumption and global trade, as agriculture continues to be a fundamental pillar of numerous economies worldwide. Guava (Psidium guajava) is a highly regarded fruit for its nutritional benefits, including high levels of dietary fiber, antioxidants, and vitamins A and C. Nevertheless, guava cultivation is encountering substantial obstacles due to various leaf diseases, which can harm the quality of the fruit and the yield. It is imperative to effectively manage these diseases to guarantee the optimal fruit production and guava crops' health. Timely and precise disease identification is critical to conventional disease management strategies, including applying fungicides, cultivating resistant varieties, and implementing cultural practices. Recent developments in deep learning, particularly Convolutional Neural Networks (CNNs), present promising solutions for the automated detection and classification of plant diseases based on visual symptoms. A CNN model that is specifically designed for the classification of guava leaf diseases, such as Canker, Dot, Mummification, Rust, and Healthy leaves, is proposed in this study. The model was trained on a comprehensive dataset that included 2108 training images and 527 testing images, and it was subsequently evaluated over a period of 30 epochs. It achieved an overall accuracy of 86.91%, with high precision, recall, and F1 scores across various disease categories. The model's potential for practical applications in agriculture, such as mobile apps or field-based systems, to assist farmers in the timely and effective management of diseases is evident in its robustness and reliability.