<![CDATA[Sugarcane is a vital crop in Indonesia, serving as the primary raw material for sugar production. Unfortunately, leaf diseases in sugarcane often pose a serious threat, potentially causing significant economic losses. These diseases are typically characterized by leaf morphological changes, making early detection and accurate classification essential to prevent further spread. This study compares three algorithms for identifying sugarcane leaf diseases: K-Nearest Neighbors (KNN), Multilayer Perceptron (MLP), and Support Vector Machine (SVM). Each algorithm employs a different approach to recognize patterns and disease characteristics: SVM separates data by identifying the optimal hyperplane, KNN classifies based on the proximity of data to training data, while MLP, as an artificial neural network, can recognize more complex patterns. The deep learning model VGG16 was utilized for feature extraction from sugarcane leaf images to enhance classification accuracy. The dataset used comprises 8,200 images of sugarcane leaves, categorized into four classes: 2,050 images of Cercospora spot gray, 2,050 of common rust, 2,050 of northern blight, and 2,050 of healthy leaves. Each category was further divided into training and testing datasets in an 80:20 ratio, with 6,560 images for training and 1,640 images for testing. The results indicate that the MLP algorithm achieved the best performance, with accuracy, precision, and recall values of 97.4%. This establishes MLP as the most effective choice for classifying sugarcane leaf diseases.]]>
Copyrights © 2024
