UCLA

Crop loss due to plant diseases and pests poses a significant challenge for crop growers worldwide, affecting product quality, nutritional value, and overall crop yield [Dir22]. Accurate disease identification is crucial for implementing effective treatments and reducing crop losses [Sol21]. This paper explores the utilization of the Vision Transformer(ViT), for image-based plant disease classification. The study builds upon the work of Mohanty et al. who used Convolutional Neural Networks to classify diseases in the PlantVillage data set [MHS16]. However, instead of Convolutional Neural Networks, this research employs a ViT model pre-trained on the ImageNet-21k data set to leverage transfer learning to train a crop disease image classification model. By developing and leveraging such tools, the agricultural community can work towards minimizing crop loss and ensuring food security for the growing global population.