Transfer Learning for Multiclass Classification of Wheat Leaf Diseases

In the following paper, we explore a study focusing on the detection of wheat leaf diseases using transfer learning techniques combined with cutting-edge convolutional neural network (CNN) architectures, namely VGG16, ResNet152V2, and Inception. The aim is to enhance the accuracy of disease detection while minimizing computational resources and training time. We conducted experiments on a dataset consisting of wheat leaf images affected with several diseases. Through transfer learning, we fine-tuned pre-trained models on our dataset to leverage their learned features. Our results demonstrate that VGG16 consistently outperforms ResNet152V2 and Inception, achieving the highest accuracy of 98.66% in disease detection. This indicates the efficacy of VGG16 for wheat leaf disease classification tasks, highlighting its potential for agricultural applications requiring robust and accurate detection of plant diseases.