Image Classification of Seasoning Package Completeness in Noodle Products Using WEKA Analysis

Klasifikasi Citra Kelengkapan Paket Bumbu Pada Produk Mie Menggunakan Analisis WEKA

Abstract

This research develops an intelligent system that utilizes vision camera technology to detect completeness of noodle packages consisting of noodle blocks, oil, and seasoning. Multi-Layer Perceptron (MLP) and Naïve Bayes are used to classify images in recognizing the shape and color of the seasoning that should be present in noodle packages using Weka. The system's input is the captured data of noodle package completeness taken in real-time with randomly positioned oil and seasoning. A total of 486 random data points were used, with 70% for training and 30% for testing. The testing results show that MLP outperforms Naïve Bayes in almost all evaluation metrics, with an accuracy of 98.48% for MLP, compared to 74.32% for Naïve Bayes. In terms of construction time, Naïve Bayes is superior with a construction time of 0.01 seconds