Identification of Potato Leaves Diseases Using CNN with Transfer Learning

Abstract

Potatoes have been among of the four main food crops grown worldwide, along with rice, sugarcane, wheat, and maize. Potatoes are a staple crop in many nations due to their excellent nutritional value. It generates more calories per unit area of land, has well-balanced protein content, and could substantially boost food production to meet population needs. Nevertheless, a variety of diseases can impact the cultivation of potatoes, resulting in substantial declines in quality and yield. The two most destructive diseases-Late blight and Early blight-have the capacity to drastically decrease the yield of potatoes. A reliable and accurate automated illness recognition system needs to be created in order to address this problem. In this study, we developed a convolution neural network model, evaluated its performance in differentiating between leaf images of the two primary diseases affecting potato plants, and contrasted the results of several CNN based models, such as Inception_ V3,Resnet_50, VGG_16, Xception and VGG19.