Klasifikasi Citra Sampah Daur Ulang Menggunakan Arsitektur MobileNetV2 dengan Strategi Fine-Tuning
DOI:
https://doi.org/10.26905/jisad.v4i1.16740Keywords:
Deep Learning;, Klasifikasi Sampah;, MobileNetV2;, Transfer Learning, Imbalance Data;Abstract
Permasalahan pengelolaan sampah menjadi isu global yang krusial, di mana pemilahan otomatis menjadi kunci efisiensi daur ulang. Penelitian ini bertujuan untuk membangun model klasifikasi sampah otomatis yang mampu mengenali lima kategori utama: kardus, logam, kertas, plastik, dan sampah residu (trash). Pendekatan yang diusulkan menggunakan Transfer Learning pada arsitektur Convolutional Neural Network (CNN) MobileNetV2 yang dikenal efisien secara komputasi. Untuk mengatasi ketidakseimbangan data (imbalance dataset), penelitian ini menerapkan teknik Class Weighting serta strategi Fine-Tuning bertahap pada lapisan base model. Model dilatih menggunakan dataset yang terdiri dari 4.272 data latih dan 2.026 data validasi. Hasil pengujian menunjukkan bahwa model mencapai akurasi sebesar 90,23% pada data validasi. Evaluasi mendalam menggunakan Confusion Matrix menunjukkan performa sangat tinggi pada kelas logam (recall 1.00) dan kertas (recall 0.96), namun menemukan tantangan signifikan pada kelas sampah residu. Hasil ini mengindikasikan bahwa meskipun MobileNetV2 sangat efektif untuk mengenali material daur ulang yang memiliki fitur bentuk tegas, variasi visual yang ekstrem pada kelas residu memerlukan penanganan khusus lebih lanjut.
Downloads
References
[1] Hartono, A. C., & Muslikh, A. R. (2025). Penerapan Transfer Learning MobileNetV2 Pada Klasifikasi Citra Jenis Buah-Buahan. Journal of Information System and Application Development, 3(2), 103-111.
[2] Kumala, R. A., Sari, C. A., & Rachmawanto, E. H. 2025. A Comparison of MobileNetV2 and VGG16 Architectures with Transfer Learning for Multi-Class Image-Based Waste Classification. Journal of Applied Informatics and Computing, 9(4).
[3] N. Walliman, Research Methods: The Basics, 3rd ed. London: Routledge, 2021. doi: 10.4324/9781003141693.
[4] Thung, G., & Yang, M. 2016. Classification of Trash for Recyclability Status. Stanford University Project Report.
[5] Yong, L., Ma, L., Sun, D., & Du, L. 2023. Application of MobileNetV2 to waste classification. PLoS One, 18(3), e0282336.
[6] Mark, S., Andrew, H., Menglong, Z., Andrey, Z., Liang-C,. MobileNetV2: Inverted Residuals and Linear Bottlenecks.
[7] A Comparison of MobileNetV2 and VGG16 Architectures with Transfer Learning for Multi-Class Image-Based Waste Classification. Journal of Applied Informatics and Computing, 9(4).
[8] Yoshua, A, S., Yuita, A, S., Sigit, A. Evaluasi Komparatif Arsitektur Lightweight CNN, MobileNetV2, dan EfficientNetB0 dalam Deteksi Penyakit Daun Jagung.
[9] Muhammad I, I., Donny, A. Squeeze-and-Excitation networks and attention mechanism in automatic detection of coffee leaf diseases based on images.
[10] Shanshan Meng, W. Chu., A Study of Garbage Classification with Convolutional Neural Networks.
[11] Sayalee Suresh Alladwar, M. Hajare, M. Scholar. Deep Learning Techniques Using Garbage Classification
[12] Q. Aini, N. Lutfiani, H. Kusumah, and M. S. Zahran, “Deteksi dan pengenalan objek dengan model machine learning: Model YOLO,” CESS (Journal of Computer Engineering System and Science), vol. 6, no. 2, Jul. 2021. [Online]. Available: https://doi.org/10.24114/cess.v6i2.25840[13] T. B. Sasongko, H. Haryoko, and A. Amrullah, “Analisis efek augmentasi dataset dan fine tune pada algoritma pre-trained convolutional neural network (CNN),” Jurnal Teknologi Informasi dan Ilmu Komputer (JTIIK), vol. 10, no. 4, pp. 763–768, Aug. 2023. [Online]. Available: https://doi.org/10.25126/jtiik.2023106583
[14] M. Sandler, A. Howard, M. Zhu, A. Zhmoginov, and L.-C. Chen, “MobileNetV2: Inverted residuals and linear bottlenecks,” in Proc. 2018 IEEE/CVF Conf. Comput. Vis. Pattern Recognit. (CVPR), 2018, pp. 4510–4520. [Online]. Available: https://doi.org/10.1109/CVPR.2018.00474
[15] U. Ozkaya, L. Seyfi. Fine-Tuning Models Comparisons on Garbage Classification for Recyclability
Downloads
Published
Issue
Section
License
Copyright (c) 2026 Journal of Information System and Application Development
This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.
Authors who publish in this journal agree to the following terms:
(1) Copyright of the published articles will be transferred to the journal as the publisher of the manuscripts. Therefore, the author confirms that the journal has managed the copyright.
(2) Publisher of JISAD: Journal of Information System and Application Development is the University of Merdeka Malang.
(3) The copyright follows Creative Commons Attribution ShareAlike License (CC BY SA): This license allows to Share - copy and redistribute the material in any medium or format, Adapt - remix, transform, and build upon the material, for any purpose, even commercially.
