Enhanced Rice Leaf Diseases Diagnosis Using Deep Learning Model with GAN-Based Synthetic Data Augmentation

Rice is a widely cultivated crop and a crucial staple food in many South Asian nations. Diseases that affect rice leaves can significantly decrease crop yields, but early identification can help mitigate their effects. Recently, machine learning has been applied to address the issue of disease spread. A major challenge, however, is the scarcity of balanced diseased leaf images compared to healthy ones, which limits the effectiveness of model training due to the need for extensive datasets. To overcome this limitation, Generative Adversarial Networks (GANs) have been increasingly used to generate synthetic images that closely mimic real ones. This approach has gained popularity in identifying leaf diseases, although research focusing specifically on rice diseases is still limited. In this research, GAN is utilized as a data augmentation method to create synthetic images for balancing Paddy doctor dataset. Subsequently, a hybrid model and Modified ResNet50 are employed for paddy disease classification. The experimental findings reveal that with image augmentation deep learning models learns well and performs best, achieving an accuracy of 99.92% by hybrid model and 98.8% by Modified ResNet50 which is 1% – 10% better when compared with existing models.