Article Per Year (5 Year)
p-Index From 2020 - 2025
0.23
P-Index
This Author published in this journals
All Journal STRING (Satuan Tulisan Riset dan Inovasi Teknologi)
Yunianto, Yunianto
Politeknik Teknologi Kimia Industri
Computer Science & IT Mathematics

Published : 1 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 1 Documents
Search

 1 
Klasifikasi Mutu Fisik Tempe Menggunakan Metode Convolutional Neural Network (CNN) Karo Karo, Ichwanul Muslim; Karo Karo, Justaman Arifin; Yunianto, Yunianto; Hariyanto, Hariyanto; Falah, Miftahul
STRING (Satuan Tulisan Riset dan Inovasi Teknologi) Vol 8, No 2 (2023)
Publisher : Universitas Indraprasta PGRI Jakarta

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.30998/string.v8i2.17596

Abstract

The quality of tempeh has until now been determined through direct physical observation. The results of observations frequently show less consistency due to human visual limitations. Image processing is an alternative used to determining the quality of tempeh from the image aspect. Image processing has capabilities that are more sensitive, precise, and objective than human vision. Convolutional Neural Network (CNN) is a deep learning model that is able to identify image objects in such a way as to determine the type of the object. In some cases, CNN algorithm is used to identify the condition of an object quality. This research aims to identify the quality of tempeh from the image aspect to ensure whether the tempeh can be classified as the tempeh having good condition or the one starting to decompose. The image of tempeh is primary data obtained directly from one of the traditional markets in Medan. The number of images that were successfully obtained was 262. The resulted classification model went through seven phases: data preparation, preprocessing, data augmentation, dataset splitting, building a classification model with the CNN algorithm with the ReLU activation function, model testing, and evaluation. The results show that the model generated from 80% of the data has an accuracy of 98.71% and a loss rate of 0.0433%. In conclusion, this study shows that the loss rate will stabilize at this rate after 50 epochs.
Co-Authors Falah, Miftahul Hariyanto, Hariyanto Ichwanul Muslim Karo Karo Karo Karo, Justaman Arifin