An Enhanced Approach for the Recognition of Rose Leaf Disease Using Vision Transformer Model

Roses are globally admired and regarded as symbols of love. It possesses considerable economic importance in agriculture. The rose blossom market is the foundation of a substantial portion of the agricultural economy. The assault of several diseases on rose plants might result in significant agronomic losses. Timely identification of these rose plant diseases enables early intervention, hence minimizing the risk of agricultural loss. Nonetheless, the traditional approach of diagnosing these disorders requires expert evaluation and is both time-intensive and expensive. This research presents an enhanced approach for detecting rose plant leaf diseases with Vision Transform (ViT-B/16) model. A total of 4342 image data has been utilized to build the model adopting various image processing techniques. The model’s performance was compared with the CNN models MobileNetV2 and ResNet50 where ViT-B/16 attained the best accuracy of 99.69%. The findings show that ViT-B/16 model has a high degree of accuracy in identifying diseases and can offer insightful information for the early diagnosis and treatment of rose plants.