Deep Learning-Based Classification of Defective Product in Images of Pedilen Feet
Bilgisayar Bilimleri ve Teknolojileri Dergisi · 03.11.2026 · 6 (2)
DOI: 10.54047/bibted.2010306 ↗Kurz gefasst
Fachartikel in der Zeitschrift Bilgisayar Bilimleri ve Teknolojileri Dergisi, online erschienen am 03.11.2026. Verfasst von Mesut Hakses und 1 weiteren Autor:innen. Die Arbeit nennt 62 Literaturangaben.
Abstract (Originalsprache)
Early and accurate identification of defective products is important for quality assurance, cost reduction, and customer satisfaction in industrial production. During Pedilen foot manufacturing, pit and bubble surface defects may occur because of insufficient chemical mixing or air trapped during injection. In this study, deep learning-based binary image classification was used to distinguish defective and non-defective products using images obtained from a Pedilen foot manufacturer. A working dataset of 1,200 images generated after image-processing/data-augmentation operations was evaluated with Xception, InceptionResNetV2, DenseNet201, NASNetMobile, and EfficientNetV2S under 10- and 20-epoch configurations. Under the existing experimental protocol, InceptionResNetV2 achieved the highest accuracy, 0.9972 (99.72%), at 20 epochs. For the same configuration, the reported precision and recall were 0.99, and the F1-score is presented as 0.99 to ensure mathematical consistency with these reported values. The findings indicate that surface defects in Pedilen foot images can be distinguished using an image-classification approach. However, because source-image-level separation before augmentation cannot be retrospectively verified and the same hold-out evaluation subset was used for model comparison, the results should be interpreted as internal comparative performance under the present dataset and experimental conditions.
Zitieren
APA
Hakses, M., & Adem, K. (2026). Deep Learning-Based Classification of Defective Product in Images of Pedilen Feet. Bilgisayar Bilimleri ve Teknolojileri Dergisi, 6(2). https://doi.org/10.54047/bibted.2010306
Vancouver
Hakses M, Adem K. Deep Learning-Based Classification of Defective Product in Images of Pedilen Feet. Bilgisayar Bilimleri ve Teknolojileri Dergisi. 2026;6(2). doi:10.54047/bibted.2010306
BibTeX
@article{Hakses2026,
author = {Hakses, Mesut and Adem, Kemal},
title = {Deep Learning-Based Classification of Defective Product in Images of Pedilen Feet},
journal = {Bilgisayar Bilimleri ve Teknolojileri Dergisi},
year = {2026},
volume = {6},
number = {2},
doi = {10.54047/bibted.2010306}
}
EUVIMED