Effect of Bilateral Filter on Support Vector Machine Algorithm for Vegetable Product Classification

Abstract

The agricultural industry is now applying artificial intelligence-based classification methods for the analysis of vegetable products. This study aims to classify vegetable products as part of the research of the classification of objects in charge that are inherently more complex than other subsets of object classification. In this research, the classification model will use the image preprocessing method on the support vector machine (SVM) algorithm. The dataset of this study amounted to 21,000 data with the division of training data (15,000 data), testing data (3,000 data) and validation data (3,000 data). In this study, experiments from the implementation of bilateral filter and support vector machine (SVM) methods obtained the highest accuracy of 70.59%. This experiment uses bilateral filters with parameters 15, 75, and 75. Other experiments obtained accuracy of 34.21% (histogram equalization) and 65.68% (colour space conversion).