Comparative Analysis of Deep Learning Models for Plant Leaf Disease Classification

Abstract

The main problem in the field of agriculture is the disease in the leaves that affects the crop and hampers the economy of farmer as well as nation. So the leaf disease classification system plays a vital role in identifying those diseases earlier so that crop production can be improved. In traditional inspection method done by a person it takes long time and requires huge resources and efforts. So due to this reason, with advancement in technologies, many advanced deep learning models and system are introduced to automate this task. In this project, we developed a basic CNN model and used other pre-trained transfer learning models such as ResNet50, VGG 16, VGG 19, AlexNet, DenseNet and EfficientNet for the purpose of leaf disease classification and compared results of all these models. The dataset we used was taken form various online resources. Finally while comparing these models, our project found that EfficientNet model and ResNet50 model were best in terms of performance and accuracy that outperformed all other models in validation, testing, and training accuracy and various other evaluation matrices such as F1 score, Recall, Precision, etc. So this project is different from other researches done in this topic and has the capacity to contribute to the overall agricultural development.