Plant Disease Classification using CNN Model and Transfer Learning

Abstract

Humans are dependent on plants in a variety of ways. Plants are expanding at a rapid rate in response to rising human and animal demands around the planet. Farmers faced numerous obstacles in cultivating their crops, including the need to preserve their plants from various illnesses, which resulted in significant financial losses. A plant is only considered healthy if it performs its biological functions to the best of its genetic capacity and so retains a similar appearance to other healthy plants. When one or more physiological activities of a healthy or normal plant are disrupted beyond a particular threshold from the usual deviation due to certain reasons (animate or inanimate stimuli; pathogens), the plant is deemed ill. Agricultural experts have been attempting for decades to develop a rapid medication system that can quickly classify plant illnesses and provide immediate therapy. In this paper, we provide a deep learning-based model for diagnosing leaf disease using the images of “potato early blight”, “potato late blight”, and “potato healthy”. Diseases cause significant losses to farmers that are interested in and involved in potato cultivation. This research uses the CNN (Convolutional Neural Network) model with transfer learning to avert these financial losses and give correct cures.