Potato Disease Detection and Classification Using Deep Learning Models

Agriculture is the backbone of food security in the world.However, crop diseases remain one of the main obstacles to yield and productivity, especially in staple crops like potatoes.Their early and accurate identification is very important for reducing damage and attaining sustainable farming.Unfortunately, the manual inspection usually practiced, or most of the prevailing machine learning models, lack accuracy and adaptability in the conditions typically found on farms.The above-mentioned challenges have been overcome in the present research by training a hybrid deep learning model, including the merits of EfficientNetV2B3 with powerful feature extraction and the Vision Transformer (ViT), one of the stateof-the-art models known for its superior attention mechanism, on a highly diversified Potato Leaf Disease Dataset (PlantVillage).Complex and realistic agricultural data can be dealt with effectively.Results achieved are very impressive -98.28%, outperforming the current state-of-the-art models by 8.6% and earlier research by 11.43%.Similarly, the values of precision, recall, and F1-score are around 0.978, which indicates very good reliability and consistency.In summary, this hybrid approach provides an overall strong, scalable, and highly accurate solution for the automated potato leaf disease detection task, marking one more step toward smarter, more sustainable agricultural practice.