Potato Leaf Disease Classification Using K-Means Cluster Segmentation and Effective Deep Learning Networks

Abstract

Potatoes are the most often consumed vegetable in many countries throughout the year, and Bangladesh is one of them. Plant diseases and venomous insects pose a significant agricultural hazard and now substantially impact Bangladesh's economy. This paper proposes a real-time technique for detecting potato leaf disease based on a deep convolutional neural network. The categorization of a picture into several categories is known as segmentation. We have used the K-means clustering algorithm for segmentation. In addition, to increase the model's efficacy, numerous data augmentation procedures have been applied to the training data. A convolutional neural network is a deep learning neural network used to prepare ordered clusters of data, such as depictions. We have used a novel CNN approach, VGG16, and ResNet50. By using VGG16, novel CNN, and resNet50, the suggested technique was able to classify potato leaves into three groups with 96, 93, and 67% accuracy, respectively. The recommended method outperforms current methodologies as we compared the performances of the models according to relevant parameters.