Image Based Pea Leaf Disease Detection with Classification Using Machine Learning Approach

Abstract

The pea is a grain crop that is consumed by both people and animals. The main protein source for under-privileged farmers. High altitude regions are where pea is produced, and for maxi mum grain production, a minimum average temperature of 30?? C is needed. Crop diseases are one of the main variables affecting pea output and productivity enhancement. One of the key methods for reducing yield production loss is the early detection of pea illnesses, although this takes a significant amount of time, money, and effort. The study suggested a machine learning method for categorizing pea illnesses based on their leaves in order to solve these issues. The design science research technique was used to achieve this. To conduct this study, a total of 2000 images were collected from Ziquala Woreda Addislem villages, Waghimra Zone, and prepared. The researcher uses picture preparation techniques after gathering the required images, such as image scaling, normalising images, and noise removal. Techniques for data augmentation were also used. It is used to identify and pick crucial characteristics that contribute to a disease's symptom. This study focuses on Ascochyta blight, downy mildew, rust, and powdery mildew as the main diseases that reduce field pea productivity. However, in Ethiopia, weed, aphids, and storage pests are also a serious productivity barrier. The image was analysed in MATLAB, and support vector machine classification was used to determine the leaf's condition. With the help of machine learning, the k-Medoid clustering technique, and the multi-SVM algorithm, this sug gested system provides an overview of the classification and detection of peat leaf illnesses using MATHLAB software and finally by inserting the pea leaf image on GUI detect, classify disease and get better accuracy result 98.3871% because instead of k-means in this work k-medoid clus tering and other statistical feature performance.