Article Per Year (5 Year)
p-Index From 2021 - 2026
0.23
P-Index
This Author published in this journals
All Journal Jurnal Saintekom : Sains, Teknologi, Komputer dan Manajemen
Sabardynata, Arjunaedy Restu
Unknown Affiliation
Computer Science & IT Decision Sciences, Operations Research & Management

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

Found 1 Documents
Search

 1 
Aplikasi Deteksi Kesegaran Ikan Menggunakan Convolutional Neural Network dan Random Forest Sabardynata, Arjunaedy Restu; Prasetyo, Eko; Tias, Rahmawati Febrifyaning
Jurnal Saintekom : Sains, Teknologi, Komputer dan Manajemen Vol 16 No 1 (2026): Maret 2026
Publisher : STMIK Palangkaraya

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.33020/saintekom.v16i1.1027

Abstract

This study is motivated by the importance of selecting healthy food in Indonesia, especially fish as a high-protein source that is highly perishable. People often struggle to distinguish fresh fish from those unfit for consumption, posing health risks. To tackle this issue, this study developed an application for detecting fish freshness using a Convolutional Neural Network (CNN) as a feature generator and Random Forest as a classifier. The CNN models employed were Residual Network 50 (Resnet50) and Visual Geometry Group 16 (VGG16). Experiments were conducted on a dataset consisting of 1,663 images of three types of fish: milkfish, tilapia, and mujair. The freshness of the fish was classified into three categories: very fresh, fresh, and not fresh. Model training utilized 80% of the data, with the remaining 20% reserved for testing. Out of a total of 333 test images (20% of the dataset), Resnet50 achieved an accuracy of 64.23% (with 86.01% accuracy for the very fresh class, 43.16% for fresh, and 52.63% for not fresh). VGG16 performed slightly better, attaining an overall accuracy of 65.16% (89.36% for very fresh, 44.90% for fresh, and 53.41% for not fresh). In terms of average accuracy, precision, recall, and F1-score, VGG16 outperforms Resnet50, although both models still make incorrect predictions. Overall, VGG16 was more effective than Resnet50 for fish freshness classification in this study.
Co-Authors Eko Prasetyo Tias, Rahmawati Febrifyaning