Developing sustainable system based on transformers algorithms to predict the Dubas insects diseases in palm leaves.

Introduction: Agriculture has emerged as a crucial area of inquiry, presenting a significant challenge for numerous experts in the field of computer vision. Identifying and categorizing plant diseases at an early stage is essential for mitigating the spread of these diseases and preventing a decline in crop yields. The overall condition of palm trees, including their roots, stems, and leaves, plays a crucial role in palm production, necessitating careful observation to ensure maximum yield. A significant challenge in maintaining productive crops is the widespread presence of pests and diseases that affect palm plants. The impact of these diseases on growth and development can be significantly negative, resulting in reduced productivity. The productivity of palms is intricately linked to the state of their leaves, which are essential for the process of photosynthesis. Methods: This study utilized an extensive dataset comprising 1600 images, which included 800 images of healthy leaves and another 800 of Dubas images. Additionally, the primary aim was to develop EfficientNetV2B0, DenseNet12, and a transformer model known as the Vision Transformer (ViT) model for detecting diseases and pests affecting palm leaves, utilizing image analysis methods to enhance pest management strategies. Results: The proposed models demonstrated superior performance compared to numerous recent studies in the field, utilizing established metrics on both original and augmented datasets, achieving an impressive accuracy of 99.37% with the ViT model. Discussion: This study presents an innovative approach for identifying diseases in palm leaves. This will have a significant impact on the agricultural sector. The results were quite promising, justifying their implementation in palm companies to improve pest and disease management.