Development Diagnosis Model of Long Bean Disease Using Image Filters with Data Mining techniques

Abstract

Growing crops, what farmers expect is produce that can be consumed and sold to make a profit. Factors that affect crop yields, including yardlong bean cultivation, are plant diseases. In order to obtain sufficient yield to meet market demand, chemicals must be used in the production process. Proper prevention and treatment of plant diseases will result in reducing damage and production costs. If farmers have tools that help analyze disease, they will be able to use chemicals correctly. The researcher therefore studied and developed a diagnostic model for yardlong bean disease by extracting image features together with data mining techniques. The diseases used in the research include powdery mildew, rust, leaf spot, leaf curl, and leaf spot. The objectives are 1) to develop a diagnostic model for yardlong bean disease by extracting image features together with data mining techniques 2) to test Performance of the developed model And in the research, image feature extraction algorithms were used, namely Simple Color Histogram Filter and Auto Color Correlogram Filter, together with data classification algorithms, namely J48, Random Forest, Random Tree and Hoeffding Tree. The results of the experiment found that the algorithms Auto Color Correlogram Filter combined with Random Forest that can diagnose disease from photographs, with an accuracy of 93.30%, precision of 94.40%, recall of 93.30% and f-measure of 93.10%, respectively, so it can be applied to develop an efficient system for diagnosing yardlong bean disease.