A Review and Challenges of Leaf Disease Prediction Using Machine Learning and Deep Learning Approach

Abstract

Even in situations of rapid population increase, agriculture provides food for everyone. It is recommended that plant diseases be anticipated early in the agricultural process in order to supply food for the whole population. On the other hand, it is regrettable to predict infections in immature crops. When examining plant diseases, a number of obstacles must be overcome, including the need for high-quality leaf images, publicly accessible datasets, noisy data influencing leaf samples, the possibility of disease identification through segmentation, but sample testing, training, and classification are also necessary. A range of diseases can be observed in different types of plants, and environmental factors can alter the color of the leaves. Various leaf diseases are categorized using machine learning (ML) and deep learning (DL) models. A workflow framework to support research in this topic is presented in this publication. While popular deep learning models for detecting leaf disease include Convolutional Neural Networks (CNN), Visual Geometry Group (VGG), ResNet (RNet), GoogLeNet, Deep CNN (DCNN), and Back Propagation Neural Networks (BPNN), popular machine learning (ML) models for predicting leaf disease include Support Vector Machine (SVM), Random Forest, and Multiple Twin SVM (MTSVM). This review would be helpful to researchers in this field who are searching for multiple efficient ML and DL-based classifiers for leaf disease detection.