Aplikasi Perangkat Uji Pupuk Berbasis Android Menggunakan Fitur Warna

Abstract

The distribution of counterfeit fertilizer has bring many disadvantages to farmers. To overcome this issue, the Ministry of Agriculture has developed the Fertilizer Test Kit that can be used directly in the field. However, the categorization of nutrient levels are still done manually by using a color chart. Therefore the nutrient level classes are limited to 2.5%, 5%, 10%, 15%, and 20%. The purpose of this research is to improve the classification of fertilizer nutrient level, both from the precision and simplicity. K-Nearest Neighbors (kNN) is used as the classification method with k = 3, 6, and 9. The histrogram of hue component in HSV color space is treated as the feature. Classification is done with 2% resolution to Nitrogen (N) and Phosphorus (P) nutrient levels. The results showed that the best accuracy is 79.54% for N with a root mean square error value (RMSE) of 1.59 and 30% for P with RMSE 4,764. Both were obtained at k = 9.