Deep Learning-Based Automated Diagnosis and Classification of Rice Leaf Diseases Using CNN Architectures for Sustainable Agriculture

A staple crop, rice feeds a substantial chunk of the world's population. However, several diseases adversely affect productivity and quality; therefore, timely diagnosis and management are required to prevent significant crop losses. Nowadays, agronomists' identification of rice leaf diseases is laborious, time-consuming, and prone to human error. Several automated and accurate disease detection and classification solutions have recently emerged with the advancements of artificial intelligence (AI) and deep learning. In this work, the results were analyzed using deep learning algorithms such as CNN, VGG19, EfficientNetB0, ResNet50, and DenseNet121 to diagnose and classify the type of rice leaf disease, including its severity and the location in which it is found. With the help of a dataset containing rice leaf images infected with bacterial leaf blight, brown spot, leaf blast, and other similar types of infection, DenseNet121 outdid all the different models and scored with an accuracy of 94, precision of 0.93, recall of 0.94 and F1-score of 0.93. The results indicate the feasibility of deep learning-based approaches to improve rice disease management practices for sustainable agricultural productivity.