EUVIMEDEUVIMED entdecken
Crossref

Impact of Image Resolution on Deep Learning-Based Classification of Elazığ Cherry Marble

Murat Yavuz, İbrahim Türkoğlu

Sinop Üniversitesi Fen Bilimleri Dergisi · 26.06.2026 · 11 (1) · 134-157

DOI: 10.33484/sinopfbd.1789939 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Sinop Üniversitesi Fen Bilimleri Dergisi, online erschienen am 26.06.2026. Verfasst von Murat Yavuz und 1 weiteren Autor:innen. Die Arbeit nennt 23 Literaturangaben. Gefördert von: This study did not receive any financial support from any public, commercial, or non-profit funding organization..

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Natural stones have played a significant role throughout human history, valued for their aesthetic, cultural and economic importance in applications ranging from monumental architecture to contemporary design. Among them, Elazığ Cherry marble, quarried exclusively in the Alacakaya district of Elazığ, Türkiye, stands out as a unique and prestigious natural resource, renowned for its deep cherry red color, distinctive vein structure and polished brilliance. This study systematically investigates the impact of image resolution on deep learning architectures for visual classification through experimental analyses conducted on Elazığ Cherry marble. A total of 2551 images were resampled into multiple resolutions ranging from 96x96 to 1024x1024 pixels and evaluated using three pre-trained architectures: ResNet50, Darknet53 and DenseNetV2. The findings demonstrate that low resolutions yielded accuracies within the 90–93% range, while intermediate resolutions (224x224 – 299x299) provided significant improvements, offering the optimal balance between accuracy and computational efficiency. At higher resolutions, performance gains became marginal; however, ResNet50 still achieved the highest accuracy of 96.10% at 1024x1024 resolution. The results highlight image resolution as a critical factor influencing not only visual quality but also classification accuracy, computational cost and model robustness against overfitting. Accordingly, this study contributes novel insights into resolution–architecture interactions and provides practical implications for the natural stone industry, delivering digital, objective and reproducible alternatives to traditional manual quality control practices.

Förderung

  • This study did not receive any financial support from any public, commercial, or non-profit funding organization.

Zitieren

APA

Yavuz, M., & Türkoğlu, İ. (2026). Impact of Image Resolution on Deep Learning-Based Classification of Elazığ Cherry Marble. Sinop Üniversitesi Fen Bilimleri Dergisi, 11(1), 134-157. https://doi.org/10.33484/sinopfbd.1789939

Vancouver

Yavuz M, Türkoğlu İ. Impact of Image Resolution on Deep Learning-Based Classification of Elazığ Cherry Marble. Sinop Üniversitesi Fen Bilimleri Dergisi. 2026;11(1):134-157. doi:10.33484/sinopfbd.1789939

BibTeX
@article{Yavuz2026,
  author = {Yavuz, Murat and Türkoğlu, İbrahim},
  title = {Impact of Image Resolution on Deep Learning-Based Classification of Elazığ Cherry Marble},
  journal = {Sinop Üniversitesi Fen Bilimleri Dergisi},
  year = {2026},
  volume = {11},
  number = {1},
  pages = {134--157},
  doi = {10.33484/sinopfbd.1789939}
}
Kostenlos starten →