Date of Publication :15th May 2024

Abstract:Traditional plant disease diagnosis methods primarily rely on visual inspection, which can be limited in accuracy and scalability. To address these challenges, this study introduces a novel framework leveraging Double GAN, a generative adversarial network, to generate synthetic diseased leaf images. This approach helps overcome imbalanced datasets commonly encountered in plant disease datasets. The synthetic images, combined with real ones, are utilized to train a deep learning model, achieving an impressive accuracy of 99.80% in disease classification tasks. Furthermore, the framework integrates recommendations for preferred pesticides based on the identified disease, enabling targeted action and potentially reducing the reliance on broad-spectrum options. This innovative approach underscores the effectiveness of deep learning coupled with data augmentation techniques for accurate plant disease detection. It also provides valuable insights for promoting sustainable crop protection practices, highlighting the potential of advanced technologies in agricultural sustainability and productivity.

Reference :