Analisis Kinerja Algoritma Machine Learning Untuk Deteksi Penyakit Daun Teh Dengan Particle Swarm Optimization

Abstract

Diseases in tea leaves are one of the causes of a decline in the quality and quantity of tea production, appropriate techniques and treatments are needed to detect diseases that can attack tea leaves. This research aims to use the best techniques to help tea farmers detect tea leaf plant diseases early. This research uses Artificial Intelligence-based techniques that apply Machine Learning algorithms to detect diseases in tea leaves. One of the challenges in implementing Machine Learning algorithms is the difficulty of finding parameters that can maximize algorithm performance. In this research, the machine learning algorithms used are Support Vector Machine (SVM) and Gradient Boosting with parameter optimization using Particle Swarm Optimization (PSO) to find the best parameters. There are 5867 images of tea leaves consisting of five types of diseases, namely Algal Spot, Brown Light, Gray Blight, Helopeltis, Red Spot, and healthy leaves used in this research. The research results show that the machine learning algorithm’s performance experienced an increase in accuracy of around 2-4% after being optimized using PSO. The accuracy obtained with the standard SVM was 87%, after optimization, it increased to 91.68%. Meanwhile, the standard Gradient Boosting obtained an accuracy of 89%, and after optimization with PSO, the accuracy increased to 91%. This research is expected to minimize the work of experts and help detect diseases on tea leaves quickly.