A Dual GAN-Based Method for Augmenting High-Quality Rice Leaf Disease Images

It is crucial to classify rice leaf diseases as accurately as possible. in the realm of agricultural administration. This paper presents a novel methodology for augmenting image data on rice leaf diseases by employing a Dual Generative Adversarial Network (GAN) framework. By utilising input image preprocessing techniques, our approach improves the dataset's quality and diversity. After that, an architecture for a Convolutional Neural Network (CNN), specifically EfficientNetB4, is utilized for categorization tasks. The evaluation metrics include accuracy, F1 score, recall, and precision. To assess if the model is effective, we compile a detailed report on classification. The creation of a convolutional neural network (CNN) model is the primary objective. that can correctly categorize paddy leaf photos into nine different disease categories: normal leaf, bacterial panicle blight, blast, hispa, tungro, and bacterial leaf blight. Spread out throughout several categories, the 10,407 tagged photos that comprise the training sample also include other information like crop variety and age. With the goal of successfully classifying each image, the suggested model is evaluated using a separate dataset that contains 3,469 photos. Our method gets a high accuracy of 99% and performs well across several evaluation metrics. This implies that it might be helpful for classifying rice leaf diseases in agricultural applications.