Journals / Bingöl Üniversitesi Teknik Bilimler Dergisi / 2026 / Sayı: 2

Comparative Analysis of Deep Learning Architectures and Performance Optimization in Mineral Image Classification

Mineral Görüntülerinin Sınıflandırılmasında Derin Öğrenme Mimarilerinin Karşılaştırmalı Analizi ve Performans Optimizasyonu

Yusuf Çelik(MUNZUR ÜNİVERSİTESİ, MÜHENDİSLİK FAKÜLTESİ)0000-0002-7859-7543
Anıl Utku(MUNZUR ÜNİVERSİTESİ, MÜHENDİSLİK FAKÜLTESİ)0000-0002-7240-8713
Pages
—
Publication Date
2026-09-25
DOI
—
🏛️Bibliographic metadata and full text access for this article are provided via DergiPark.
DergiPark Record ↗
✨

AI Summary & Key Takeaways

AI Insight
⚡ TL;DR(Too Long; Didn't Read)

The findings indicate that convolutional neural networks are more effective than transformer-based models in the mineral image classification problem.

📌 3 Key Takeaways

🎯
Objective & Hypothesis:Accurate and rapid classification of minerals is of great importance in geology, mining, and industrial applications.
📊
Key Findings & Data:In this study, a deep learning-based approach is proposed for the automatic classification of mineral images, and the performance of different model architectures is analyzed comparatively.
💡
Conclusion & Impact:The findings indicate that convolutional neural networks are more effective than transformer-based models in the mineral image classification problem.
🔬Design:Modeling & Simulation
📍Scope:Comprehensive Sample
⚙️Variables:Mineral sınıflandırma · derin öğrenme · görüntü işleme

Abstract

Accurate and rapid classification of minerals is of great importance in geology, mining, and industrial applications. Traditional methods are often based on expert knowledge and experience, and can produce time-consuming and subjective results. In this study, a deep learning-based approach is proposed for the automatic classification of mineral images, and the performance of different model architectures is analyzed comparatively. The Mineral Photos dataset, containing a total of 40,354 images belonging to 15 different mineral classes, was used in the study. In the experimental process, five different pre-trained models were evaluated: ResNet50, EfficientNet-B0, MobileNetV3-Large, DenseNet121, and Vision Transformer (ViT-B/16). All models were tested using a 5-fold cross-validation method under the same training protocol. According to the first stage results, the EfficientNet-B0 model achieved the highest performance with 74.39% accuracy, while the ResNet50 (71.37%) and DenseNet121 (70.13%) models presented competitive results. In contrast, the ViT-B/16 model showed lower performance compared to other models with an accuracy of 48.61%. In the second stage, more advanced experiments were conducted on EfficientNet-based models, and it was determined that the EfficientNet-V2-S model exhibited higher and more stable performance with an average accuracy of 83.14%. Furthermore, the model produced balanced results in precision, recall, and F1-score metrics. The findings indicate that convolutional neural networks are more effective than transformer-based models in the mineral image classification problem.

Özet

Minerallerin doğru ve hızlı bir şekilde sınıflandırılması; jeoloji, madencilik ve endüstriyel uygulamalar açısından büyük önem taşımaktadır. Geleneksel yöntemler çoğunlukla uzman bilgi ve deneyimine dayalı olup zaman alıcı ve öznel sonuçlar üretebilmektedir. Bu çalışmada, mineral görüntülerinin otomatik sınıflandırılması amacıyla derin öğrenme tabanlı bir yaklaşım önerilmiş ve farklı model mimarilerinin performansları karşılaştırmalı olarak analiz edilmiştir. Çalışmada, 15 farklı mineral sınıfına ait toplam 40.354 görüntü içeren Mineral Photos veri seti kullanılmıştır. Deneysel süreçte, ResNet50, EfficientNet-B0, MobileNetV3-Large, DenseNet121 ve Vision Transformer (ViT-B/16) olmak üzere beş farklı önceden eğitilmiş model değerlendirilmiştir. Tüm modeller aynı eğitim protokolü altında 5-katlı çapraz doğrulama yöntemi ile test edilmiştir. İlk aşama sonuçlarına göre EfficientNet-B0 modeli %74.39 doğruluk ile en yüksek performansı elde ederken, ResNet50 (%71.37) ve DenseNet121 (%70.13) modelleri rekabetçi sonuçlar sunmuştur. Buna karşılık ViT-B/16 modeli %48.61 doğruluk ile diğer modellere kıyasla daha düşük performans göstermiştir. İkinci aşamada, EfficientNet tabanlı modeller üzerinde daha gelişmiş deneyler gerçekleştirilmiş ve EfficientNet-V2-S modelinin %83.14 ortalama doğruluk değeri ile daha yüksek ve daha kararlı bir performans sergilediği belirlenmiştir. Ayrıca model, precision, recall ve F1-skoru metriklerinde de dengeli sonuçlar üretmiştir. Elde edilen bulgular, konvolüsyonel sinir ağlarının mineral görüntü sınıflandırma probleminde transformer tabanlı modellere kıyasla daha etkili olduğunu göstermektedir.

🔖

Cite This Article

Çelik, Y., Utku, A. (2026). Comparative Analysis of Deep Learning Architectures and Performance Optimization in Mineral Image Classification. Bingöl Üniversitesi Teknik Bilimler Dergisi, Sayı: 2. https://ajindex.com/articles/1610660

📄 Full Text PDF Viewer

📄

PDF görüntüleyici hazırlanıyor...

References (38)

  1. Tuğrul, A., & Yılmaz, M. (2023). Agrega Üretim Sahalarındaki Belirsizlik ve Risklerin Azaltılmasında Jeoloji ve Mühendislik Jeolojisi Araştırmaların Önemi. Jeoloji Mühendisliği Dergisi, 47(1), 87-102.
  2. Balaram, V. (2023). Advances in analytical techniques and applications in exploration, mining, extraction, and metallurgical studies of rare earth elements. Minerals, 13(8), 1031.
  3. Balaram, V., & Sawant, S. S. (2022). Indicator minerals, pathfinder elements, and portable analytical instruments in mineral exploration studies. Minerals, 12(4), 394.
  4. Long, T., Zhou, Z., Hancke, G., Bai, Y., & Gao, Q. (2022). A review of artificial intelligence technologies in mineral identification: classification and visualization. Journal of Sensor and Actuator Networks, 11(3), 50.
  5. Ling, Q., & Isa, N. A. M. (2023). Printed circuit board defect detection methods based on image processing, machine learning and deep learning: A survey. IEEE access, 11, 15921-15944.
  6. Taye, M. M. (2023). Theoretical understanding of convolutional neural network: Concepts, architectures, applications, future directions. Computation, 11(3), 52.
  7. Wang, Y., Deng, Y., Zheng, Y., Chattopadhyay, P., & Wang, L. (2025). Vision transformers for image classification: A comparative survey. Technologies, 13(1), 32.
  8. Tsangaratos, P., Ilia, I., Spanoudakis, N., Karageorgiou, G., & Perraki, M. (2025). Machine learning approaches for real-time mineral classification and educational applications. Applied Sciences, 15(4), 1871.
  9. Chawshin, K., Berg, C. F., Varagnolo, D., & Lopez, O. (2021). Lithology classification of whole core CT scans using convolutional neural networks. SN Applied Sciences, 3(6), 668.
  10. Xu, Z., Ma, W., Lin, P., & Hua, Y. (2022). Deep learning of rock microscopic images for intelligent lithology identification: Neural network comparison and selection. Journal of Rock Mechanics and Geotechnical Engineering, 14(4), 1140-1152.
  11. Zhou, W., Wang, H., & Wan, Z. (2022). Ore image classification based on improved CNN. Computers and Electrical Engineering, 99, 107819.
  12. Zhou, Z., Yuan, H., & Cai, X. (2023). Rock thin section image identification based on convolutional neural networks of adaptive and second-order pooling methods. Mathematics, 11(5), 1245.
  13. Rizaldy, A., Afifi, A. J., Ghamisi, P., & Gloaguen, R. (2024). Improving mineral classification using multimodal hyperspectral point cloud data and multi-stream neural network. Remote Sensing, 16(13), 2336.
  14. Li, S., Dong, Y., Zhang, Z., Lin, C., Liu, H., Wang, Y.,. & Zhang, G. (2024). Study on automatic lithology identification based on convolutional neural network and deep transfer learning. Discover Applied Sciences, 6(6), 322.
  15. Ma, W., Han, T., Xu, Z., & Lin, P. (2025). Feature fusion of single and orthogonal polarized rock images for intelligent lithology identification. AI in Civil Engineering, 4(1), 5.
  16. Wang, M., Guo, W., Yang, F., Yan, B., Xu, Y., Jiang, J., & Huang, J. (2025). Rock thin section image classification in low data scenarios using few-shot learning. Computers & Geosciences, 203, 105962.
  17. Geillon, F. Mineral Photos Veri Seti. Kaggle. Erişim adresi: https://www.kaggle.com/datasets/floriangeillon/mineral-photos
  18. Zhang, L., Bian, Y., Jiang, P., & Zhang, F. (2023). A transfer residual neural network based on ResNet-50 for detection of steel surface defects. Applied Sciences, 13(9), 5260.
  19. Wen, L., Li, X., & Gao, L. (2020). A transfer convolutional neural network for fault diagnosis based on ResNet-50. Neural Computing and Applications, 32(10), 6111-6124.
  20. Ali, R. R., Yaacob, N. M., Alqaryouti, M. H., Sadeq, A. E., Doheir, M., Iqtait, M.,. & Yaacob, S. S. (2025). Learning architecture for brain tumor classification based on deep convolutional neural network: Classic and ResNet50. Diagnostics, 15(5), 624.
  21. Padhi, J., Korada, L., Dash, A., Sethy, P. K., Behera, S. K., & Nanthaamornphong, A. (2024). Paddy leaf disease classification using EfficientNet B4 with compound scaling and swish activation: A deep learning approach. IEEE Access, 12, 126426-126437.
  22. Ali, H., Shifa, N., Benlamri, R., Farooque, A. A., & Yaqub, R. (2025). A fine tuned EfficientNet-B0 convolutional neural network for accurate and efficient classification of apple leaf diseases. Scientific Reports, 15(1), 25732.
  23. Zhang, J., Yang, X., Fu, X., Wang, B., & Li, H. (2025). LDL-MobileNetV3S: an enhanced lightweight MobileNetV3-small model for potato leaf disease diagnosis through multi-module fusion. Frontiers in Plant Science, 16, 1656731.
  24. Duan, Z., & Liu, H. (2025). Improving Network Structure for Efficient Classification Network Based on MobileNetV3. IEEE Access, 13, 191296-191308.
  25. Yehulu, G. T., Walle, Y. M., Haile, M. B., Yohannes, S. T., Tegegne, A. G., Tesfaye, D. G., & Baye, G. A. (2025). Mango fruit disease detection and classification using MobileNetV3_large model. Scientific African, e03061.
  26. Albelwi, S. A. (2022). Deep architecture based on DenseNet-121 model for weather image recognition. International Journal of Advanced Computer Science and Applications, 13(10).
  27. Prakash, N. N., Rajesh, V., Namakhwa, D. L., Pande, S. D., & Ahammad, S. H. (2023). A DenseNet CNN-based liver lesion prediction and classification for future medical diagnosis. Scientific African, 20, e01629.
  28. Hong, S., Wu, J., Zhu, L., & Chen, W. (2024). Brain tumor classification in VIT-B/16 based on relative position encoding and residual MLP. Plos one, 19(7), e0298102.
  29. Appati, J. K., Ziamah, B., Akrofi, H. A., & Dodoo, A. A. (2025). SARS detection in chest CT scan images using the bootstrapped ViT-B/16 model. Iran Journal of Computer Science, 8(2), 537-551.
  30. Kumar, P., Mathew, J., & Sanodiya, R. K. (2025). Optimized loss and self attention for enhanced domain adaptation in remote sensing image classification. Scientific Reports, 15(1), 40259.
  31. Ganesh, P., Chen, Y., Lou, X., Khan, M. A., Yang, Y., Sajjad, H.,. & Winslett, M. (2021). Compressing large-scale transformer-based models: A case study on bert. Transactions of the Association for Computational Linguistics, 9, 1061-1080.
  32. Vujović, Ž. (2021). Classification model evaluation metrics. International Journal of Advanced Computer Science and Applications, 12(6), 599-606.
  33. Cabot, J. H., & Ross, E. G. (2023). Evaluating prediction model performance. Surgery, 174(3), 723-726.
  34. Foody, G. M. (2023). Challenges in the real world use of classification accuracy metrics: From recall and precision to the Matthews correlation coefficient. Plos one, 18(10), e0291908.
  35. Obi, J. C. (2023). A comparative study of several classification metrics and their performances on data. World Journal of Advanced Engineering Technology and Sciences, 8(1), 308-314.
  36. Li, S., Dong, Y., Zhang, Z., Lin, C., Liu, H., Wang, Y.,. & Zhang, G. (2024). Study on automatic lithology identification based on convolutional neural network and deep transfer learning. Discover Applied Sciences, 6(6), 322.
  37. Ali, R. R., Yaacob, N. M., Alqaryouti, M. H., Sadeq, A. E., Doheir, M., Iqtait, M.,. & Yaacob, S. S. (2025). Learning architecture for brain tumor classification based on deep convolutional neural network: Classic and ResNet50. Diagnostics, 15(5), 624.
  38. Ganesh, P., Chen, Y., Lou, X., Khan, M. A., Yang, Y., Sajjad, H.,. & Winslett, M. (2021). Compressing large-scale transformer-based models: A case study on bert. Transactions of the Association for Computational Linguistics, 9, 1061-1080.