Detection and Classification of Plant Leaf Diseases Using Digtal Image Processing Methods: A Review

Abstract

Prediction and classification of plant leaf illnesses by farmers using conventional approaches can be unexciting and erroneous. Problems may occur while trying to predict the sort of illnesses manually. Inability to detect diseases of plants promptly could lead to the destruction of crop plants and this can cause serious decline in the yield. Losses can be prevented and yield be maximized when farmers adopt computerized image processing approaches in their farms. Numerous techniques have been proposed and used in the prediction of diseases of crop plants based on the images of the infected leaves. Researchers have in the past achieved a lot in the aspect of plant illnesses identification by exploring several techniques and models. However, improvement needs to be provided on account of reviews, new advancements and discussions. Deploying technology can greatly enhance crop production across the globe. Different approaches and models can be trained with huge data to identify new improved methods for uncovering diseases of plants to tackle problem of low yield. Previous works have determined the robustness of various image processing techniques such as; K-means clustering, Naive Bayes, Feed forward neutral network (FFNN), Support Vector Machine (SVM), K-nearest neighbor (KNN) classifier, Fuzzy logic, Genetic Algorithm (GA), Artificial Neural Network (ANN), Convolutional Neural Network (CNN) etc. This paper provides a critical review and results of different types of approaches and methods used previously to detect and classify various types of plant leaf illnesses using image processing approaches.

Leaf disease severity classification with explainable artificial intelligence using transformer networks

“…Mohammed and Yusoff [41] provides an overview of various techniques available for achieving success in machine learning, deep learning, and image processing. It emphasizes the importance of training and testing models with more datasets to increase the accuracy of recognition rates.…”