Published Paper Details:

Agribusiness serves as the primary livelihood for approximately 70% of rural residents, with India ranking as the world's second-largest producer of various agricultural commodities. The agriculture industry significantly impacts India's gross domestic product (GDP). Technological advancements play a significant influence in enhancing agricultural practices, enabling the prediction of considerations include soil quality, crop health, and disease detection to optimize crop yields.Detecting and addressing plant leaf diseases early is paramount to prevent their spread and enhance overall yield. This research focuses on plant leaf disease detection and stage classification based on infection severity. The approach involves leveraging deep learning models, specifically YOLOv5 for disease detection, followed by background removal using the U2-Net architecture. Subsequently, a vision transformer (ViT) is employed for stage classification, categorizing severity into low, moderate, and high.To combat leaf diseases effectively, a recommendation solution is proposed. YOLOv5 is trained using open-source datasets, namely PlantDoc and Plantvillage. This study predominantly concentrates on apple leaf disease classification. Results indicate that YOLOv5 achieves a maximum F1-score of 0.57 at a confidence score of 0.2, while the vision transformer, when applied with and without background images, attains F1-scores of 0.758 and 0.908, respectively.