Implementasi Support Vector Machine (SVM) pada Klasifikasi Jenis Tanah Memanfaatkan Fitur RGB

Abstract

Each type of soil possesses unique characteristics that influence various aspects of human life and activities, particularly in agriculture, construction, and ecology. Naturally, soil exhibits distinct colors that aid human interpretation in identifying soil types. However, these soil types are not widely known. Understanding soil types can facilitate better decision-making across various aspects of life and contribute to environmental conservation and natural resource management. Given the diversity of soil types with varying properties, tailored treatments for each type are essential. Therefore, soil classification is crucial for understanding effective soil management practices. The Support Vector Machine (SVM) algorithm excels in handling high-dimensional data, separating non-linear classes, and enhancing model accuracy and generalization, making it a robust choice for soil classification based on RGB (Red, Green, Blue) features. This research method involved collecting a personal dataset comprising various soil types, extracting RGB color features, and using first-order statistics for data representation. The results demonstrate that the optimized combination of RGB features and SVM achieved accurate and efficient soil classification with an accuracy of 88%.