DiaMOS Plant Leaves Disease Classification using Vision Transformer
C Sugunadevi, Rimjhim Padam Singh, B. Uma Maheswari, Priyanka Kumar
Abstract
The consequences of climate change have increased during the past several years. This climate influences each stage of plant production, and farmers are being forced to modify the plantation and advance their crop management techniques utilizing recent technology built on data analytics. Many plant diseases manifest themselves in the leaves. Therefore, successful disease diagnosis requires a thorough understanding of the condition of plants. To provide high-quality agricultural products, timely and precise disease detection and classification are crucial for smart and precision agriculture. In this study, images of the DiaMOS plant leaves dataset that was collected in the field of Sardegna, Italy is utilized to learn and classify plant disease using several fine-tuned z deep learning models like InceptionV3, EfficientNetB0, VGG-16, ResNet50, VGG-19. Here, the Deep learning models are trained to categorize the severity of plant disease and built a vision transformer for classification that achieves higher than the predefined models with an accuracy- 98.5% precision-97.1%, and F1 score – 97.83%.