A Comparative Study of MobileNetV2 and ResNet50 for Multi-Class AI- Generated and Real Image Classification

Authors

  • I Gusti Ngurah Agus Ega Patria Pramudita Informatika, Fakultas Teknologi dan Informatika, Institut Bisnis dan Teknologi Indonesia, Denpasar, Bali, Indonesia
  • I Gede Iwan Sudipa Informatika, Fakultas Teknologi dan Informatika, Institut Bisnis dan Teknologi Indonesia, Denpasar, Bali, Indonesia
  • Yuri Prima Fittryani Informatika, Fakultas Teknologi dan Informatika, Institut Bisnis dan Teknologi Indonesia, Denpasar, Bali, Indonesia
  • Ida Bagus Ary Indra Iswara Informatika, Fakultas Teknologi dan Informatika, Institut Bisnis dan Teknologi Indonesia, Denpasar, Bali, Indonesia
  • I Gusti Ayu Agung Mas Aristamy Informatika, Fakultas Teknologi dan Informatika, Institut Bisnis dan Teknologi Indonesia, Denpasar, Bali, Indonesia

DOI:

10.33395/sinkron.v10i1.15682

Keywords:

AI-Generated Images, Real Images, Deep Learning, Convolutional Neural Network (CNN), Image Classification

Abstract

This study aims to classify AI-generated and real images using Convolutional Neural Network (CNN) architecture by comparing the performance of MobileNetV2 and ResNet50. Previous studies on AI-generated image detection have primarily focused on binary classification without explicitly analyzing object-level context in multi-class scenarios, leaving a gap in understanding model performance across diverse visual categories. The dataset consists of 23,941 images divided into two main classes of real and fake and five subclasses of human, animal, art, view, and vehicle. The training process employs data augmentation and a K-Fold Cross Validation strategy on the training and validation set to maintain balanced class proportions, while a separate unseen test set is used exclusively for final performance evaluation. Model evaluation is performed based on accuracy, precision, recall, and F1-score metrics on test data. The results showed that MobileNetV2 achieved the best accuracy of 89% at the 10th epoch, but experienced a decline in performance at the 30th and 50th epochs, indicating overfitting. In contrast, ResNet50 showed the most stable performance with the highest accuracy of 93% at the 30th epoch and consistently high precision, recall, and F1-score values. Thus, ResNet50 was found to be the most effective architecture for classification of AI-generated and real images on multi-class datasets, while MobileNetV2 remains relevant for implementation on devices with computational limitations.

GS Cited Analysis

Downloads

Download data is not yet available.

References

Aljohani, K., & Turki, T. (2022). Automatic Classification of Melanoma Skin Cancer with Deep Convolutional Neural Networks. AI (Switzerland), 3(2), 512–525. https://doi.org/10.3390/ai3020029

Bahrul Subkhi, M., Bagus Setiawan, A., & Yusuf Alif Candra, M. (2023). Klasifikasi Gambar: Membedakan Lukisan Buatan Manusia dan AI dengan CNN. Jurnal Filsafat, Sains, Teknologi, Dan Sosial Budaya, 29(4), 149–155. https://doi.org/10.33503/paradigma.v30i4.1284

Bichri, H., Chergui, A., & Hain, M. (2024). Investigating the Impact of Train / Test Split Ratio on the Performance of Pre-Trained Models with Custom Datasets. IJACSA) International Journal of Advanced Computer Science and Applications, 15(2). https://doi.org/10.14569/IJACSA.2024.0150235

Daviana, F. P., Aryanti, A., & Anugraha, N. (2025). Performance Comparison Between ResNet50 and MobileNetV2 for Indonesian Sign Language Classification. JURIKOM (Jurnal Riset Komputer), 12(3), 319–328. https://doi.org/10.30865/jurikom.v12i3.8667

de Oro, J. E. C. G., Koch, P. J., Krois, J., Ros, A. G. C., Patel, J., Meyer-Lueckel, H., & Schwendicke, F. (2022). Hyperparameter Tuning and Automatic Image Augmentation for Deep Learning-Based Angle Classification on Intraoral Photographs—A Retrospective Study. Diagnostics, 12(7). https://doi.org/10.3390/diagnostics12071526

Döring, N., Le, T. D., Vowels, L. M., Vowels, M. J., & Marcantonio, T. L. (2024). The Impact of Artificial Intelligence on Human Sexuality: A Five-Year Literature Review 2020–2024. Current Sexual Health Reports, 17(1), 4. https://doi.org/10.1007/s11930-024-00397-y

Fatoni, Kurniawan, T. B., Dewi, D. A., Zakaria, M. Z., & Muhayeddin, A. M. M. (2025). Fake vs Real Image Detection Using Deep Learning Algorithm. Journal of Applied Data Sciences, 6(1), 366–376. https://doi.org/10.47738/jads.v6i1.490

Ghiurău, D., & Popescu, D. E. (2025). Distinguishing Reality from AI: Approaches for Detecting Synthetic Content. Computers, 14(1), 1–33. https://doi.org/10.3390/computers14010001

Hajar, S., Murinto, M., & Yudhana, A. (2025). Comparison of Transfer Learning Strategies Using MobileNetV2 and ResNet50 for Ecoprint Leaf Classification. Jurnal Teknik Informatika (Jutif), 6(5), 3251–3264. https://doi.org/10.52436/1.jutif.2025.6.5.5266

Hakim, S. A., Ubaidillah, M., Ramadhan, A. R., Zulvia, R., Hawari, A., Rizky, A. B., Lutfi, R., Tsania, P., Hermanto, M., Yudistira, N., & Korespondensi, P. (2024). KLASIFIKASI CITRA GENERASI ARTIFICIAL INTELLIGENCE MENGGUNAKAN METODE FINE TUNING PADA RESIDUAL NETWORK. Jurnal Teknologi Informasi Dan Ilmu Komputer (JTIIK), 11(3), 655–666. https://doi.org/10.25126/jtiik.938118

Hang Rai, D. (2024). Artificial Intelligence Through Time: A Comprehensive Historical Review. Bachelor of Science in Computer Science and Information Technology. https://doi.org/10.13140/RG.2.2.22835.03364

Kanza, S., & Knight, N. J. (2022). Behind every great research project is great data management. In BMC Research Notes (Vol. 15, Issue 1). BioMed Central Ltd. https://doi.org/10.1186/s13104-022-05908-5

Komendantova, N., & Erokhin, D. (2025). Artificial Intelligence Tools in Misinformation Management during Natural Disasters. Public Organization Review, 1–25. https://doi.org/10.1007/s11115-025-00815-2

Kumar, T., Mileo, A., Brennan, R., & Bendechache, M. (2023). Image Data Augmentation Approaches: A Comprehensive Survey and Future directions. Science Foundation Ireland. https://doi.org/10.48550/arXiv.2301.02830

Li, Z., Liu, F., Yang, W., Peng, S., & Zhou, J. (2022). A Survey of Convolutional Neural Networks: Analysis, Applications, and Prospects. IEEE Transactions on Neural Networks and Learning Systems, 33(12), 6999–7019. https://doi.org/10.1109/TNNLS.2021.3084827

Lin, C. C., Huang, A. Y. Q., & Lu, O. H. T. (2023). Artificial intelligence in intelligent tutoring systems toward sustainable education: a systematic review. In Smart Learning Environments (Vol. 10, Issue 1). Springer. https://doi.org/10.1186/s40561-023-00260-y

Mu, J., Adrezo, M., & Haikal, A. N. (2024). Identifikasi Wajah Asli dan Buatan Deepfake Menggunakan Metode Convolutional Neural Network. TEKNIKA, 13(1), 45–50. https://doi.org/10.34148/teknika.v13i1.705

Peng, Y. (2024). A Comparative Analysis Between GAN and Diffusion Models in Image Generation. Transactions on Computer Science and Intelligent Systems Research, 5, 189–195. https://doi.org/10.62051/0f1va465

Pradnya Dhuhita, W. M., Ubaid, M. Y., & Baita, A. (2023). MobileNet V2 Implementation in Skin Cancer Detection. ILKOM Jurnal Ilmiah, 15(3), 498–506. https://doi.org/10.33096/ilkom.v15i3.1702.498-506

Słapczyński, T. (2022). Artificial Intelligence in Science and Everyday Life, Its Application and Development Prospects. ASEJ - Scientific Journal of Bielsko-Biala School of Finance and Law, 26(4), 78–85. https://doi.org/10.19192/wsfip.sj4.2022.12

Sophia LI. (2025). The Social Harms of AI-Generated Fake News: Addressing Deepfake and AI Political Manipulation. Digital Society & Virtual Governance, 1(1), 72–88. https://doi.org/10.6914/dsvg.010105

Suharyanto, D., Lubis, C., & Dharmawan, A. B. (2024). PENERAPAN CONVOLUTIONAL NEURAL NETWORK DAN CAPSULE NETWORKS DALAM MENDETEKSI DEEPFAKE. Jurnal Ilmu Komputer Dan Sistem Informasi, 12(1). https://doi.org/10.24912/jiksi.v12i1.28190

Sun, Y., Miao, L., Zhao, Z., Pan, T., Wang, X., Guo, Y., Xin, D., Chen, Q., & Zhu, R. (2023). An Efficient and Automated Image Preprocessing Using Semantic Segmentation for Improving the 3D Reconstruction of Soybean Plants at the Vegetative Stage. Agronomy, 13(9). https://doi.org/10.3390/agronomy13092388

Yun, S., Oh, S. J., Heo, B., Han, D., Choe, J., & Chun, S. (2021). Re-labeling ImageNet: from Single to Multi-Labels, from Global to Localized Labels. CVPR. https://doi.org/10.48550/arXiv.2101.05022

Downloads


Crossmark Updates

How to Cite

Pramudita, I. G. N. A. E. P., Sudipa, I. G. I., Fittryani, Y. P. ., Iswara, I. B. A. I. ., & Aristamy, I. G. A. A. M. . (2026). A Comparative Study of MobileNetV2 and ResNet50 for Multi-Class AI- Generated and Real Image Classification. Sinkron : Jurnal Dan Penelitian Teknik Informatika, 10(1), 536-549. https://doi.org/10.33395/sinkron.v10i1.15682

Issue

Vol. 10 No. 1 (2026): Article Research January 2026

Section

Articles

License

Copyright (c) 2026 I Gusti Ngurah Agus Ega Patria Pramudita, I Gede Iwan Sudipa, Yuri Prima Fittryani, Ida Bagus Ary Indra Iswara, I Gusti Ayu Agung Mas Aristamy

Creative Commons License

This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.

Crossref
Scopus
Google Scholar
Europe PMC

Most read articles by the same author(s)

1 2 > >>