Journals / Zeki Sistemler Teori ve Uygulamaları Dergisi / 2026 / Cilt: 9 Sayı: 2026

Comparison of Deep Learning Architectures for Fabric Defect Detection in Textile Production Lines and a New Image Dataset

Tekstil Üretim Hatlarında Kumaş Hata Tespiti İçin Derin Öğrenme Mimarilerinin Karşılaştırılması ve Yeni Bir Görüntü Veri Seti

Yavuz Canbay(Kahramanmaraş Sütçü İmam Üniversitesi, Mühendislik ve Mimarlık Fakültesi, Bilgisayar Mühendisliği Bölümü)0000-0003-2316-7893
Mervecan Cengiz(KAHRAMANMARAS SUTCU IMAM UNIVERSITY)0009-0002-9932-2589
Ahmet Talha Sivrikaya(KAHRAMANMARAŞ SÜTÇÜ İMAM ÜNİVERSİTESİ)0009-0008-8337-5524
Pages
1–15
Publication Date
2026-10-28
🏛️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)

Experimental results demonstrated that the MobileNetV2 model achieved the highest numerical performance values—99.31% accuracy, 99.32% precision, 99.31% recall, and a 99.31% F1-score—using the existing fixed data split.

📌 3 Key Takeaways

🎯
Objective & Hypothesis:Quality control processes in the textile industry rely heavily on human inspection, which results in significant limitations such as high costs, low speed, and the risk of human error.
📊
Key Findings & Data:In this study, an automated fabric defect assessment system based on image-level binary classification was developed, aiming to separate fabric images obtained from textile production lines into defective and defect-free categories.
💡
Conclusion & Impact:Experimental results demonstrated that the MobileNetV2 model achieved the highest numerical performance values—99.31% accuracy, 99.32% precision, 99.31% recall, and a 99.31% F1-score—using the existing fixed data split.
🔬Design:Modeling & Simulation
📍Scope:Comprehensive Sample
⚙️Variables:Kumaş kusur tespiti · Derin öğrenme · Endüstriyel görüntü işleme

Abstract

Quality control processes in the textile industry rely heavily on human inspection, which results in significant limitations such as high costs, low speed, and the risk of human error. In particular, it is difficult to detect fabric defects at the millimeter level manually, which can lead to defective products passing through the production line. In this study, an automated fabric defect assessment system based on image-level binary classification was developed, aiming to separate fabric images obtained from textile production lines into defective and defect-free categories. A new dataset consisting of 2,915 images with a resolution of 512×512 was created. Twenty percent of the dataset was reserved as an independent test set, while 10% of the remaining 80% development set was used for validation. Five different deep learning architectures, namely Original CNN, VGG16, ResNet50, InceptionV3, and MobileNetV2, were comparatively analyzed. Experimental results demonstrated that the MobileNetV2 model achieved the highest numerical performance values—99.31% accuracy, 99.32% precision, 99.31% recall, and a 99.31% F1-score—using the existing fixed data split. Furthermore, MobileNetV2's low-parameter and computationally lightweight architecture offers a potential advantage for real-time industrial applications. However, real-time operational performance was not verified in this study using hardware-specific FPS or latency measurements. The findings reveal that lightweight and optimized deep learning architectures can successfully achieve a balance between performance and complexity in the textile industry.

Özet

Tekstil endüstrisinde kalite kontrol süreçleri büyük ölçüde insan denetimine dayanmaktadır ve bu durum yüksek maliyet, düşük hız ve insan kaynaklı hata riski gibi önemli sınırlamalar doğurmaktadır. Özellikle milimetrik boyuttaki kumaş kusurlarının manuel olarak tespit edilmesi zor olup, hatalı ürünlerin üretim hattından geçmesine neden olabilmektedir. Bu çalışmada, tekstil üretim hatlarından elde edilen kumaş görüntülerinin hatalı veya hatasız olarak ayrılmasını amaçlayan, görüntü düzeyinde ikili sınıflandırmaya dayalı otomatik bir kumaş kusur değerlendirme sistemi geliştirilmiştir. Çalışma kapsamında 512×512 çözünürlüğünde 2915 görüntüden oluşan yeni bir veri seti oluşturulmuştur. Veri setinin %20'si bağımsız test kümesi olarak ayrılmış, kalan %80'lik geliştirme kümesinin %10'u doğrulama amacıyla kullanılmıştır. Özgün CNN, VGG16, ResNet50, InceptionV3 ve MobileNetV2 olmak üzere beş farklı derin öğrenme mimarisi karşılaştırmalı olarak analiz edilmiştir. Deneysel sonuçlar, MobileNetV2 modelinin mevcut sabit veri bölünmesinde %99,31 doğruluk, %99,32 kesinlik, %99,31 duyarlılık ve %99,31 F1-skor ile sayısal olarak en yüksek performans değerlerini elde ettiğini göstermiştir. Ayrıca MobileNetV2’nin düşük parametreli ve hesaplama açısından hafif mimarisi, gerçek zamanlı endüstriyel uygulamalar açısından potansiyel bir avantaj sunmaktadır. Bununla birlikte, gerçek zamanlı çalışma performansı bu çalışmada donanıma özgü FPS veya gecikme ölçümleriyle doğrulanmamıştır. Elde edilen bulgular, hafif ve optimize edilmiş derin öğrenme mimarilerinin tekstil sektöründe performans ve karmaşıklık dengesini başarıyla sağlayabileceğini ortaya koymaktadır.

🔖

Cite This Article

Canbay, Y., Utku, A., Cengiz, M., Sivrikaya, A. T. (2026). Comparison of Deep Learning Architectures for Fabric Defect Detection in Textile Production Lines and a New Image Dataset. Zeki Sistemler Teori ve Uygulamaları Dergisi, Cilt: 9 Sayı: 2026, 1–15. https://doi.org/10.38016/jista.1933403

📄 Full Text PDF Viewer

📄

PDF görüntüleyici hazırlanıyor...

References (33)

  1. Alruwais, N., Alabdulkreem, E., Mahmood, K., Marzouk, R., Assiri, M., Abdelmageed, A. A., Abdelbagi, S., & Drar, S., 2023. Hybrid mutation moth flame optimization with deep learning-based smart fabric defect detection. Computers and Electrical Engineering, 108, 108706.
  2. Arshad, S. R., & Shahzad, M. K., 2024. Deep Learning Based Fabric Defect Detection. Research Reports on Computer Science, 1-11.
  3. Beljadid, A., Tannouche, A., & Balouki, A., 2022. Automatic fabric defect detection employing deep learning. International Journal of Electrical and Computer Engineering (IJECE), 12(4), 4129.
  4. Canbay, Y., & Selver, E., 2022. Kumaş Hatası Tespitinde Ön Eğitimli Derin Öğrenme Modellerinin Performanslarının Karşılaştırılması. V. Internatıonal Conference on Data Science and Applications, September 7-11.
  5. Canbay, Y., Adsiz, S., & Canbay, P., 2024. Privacy-preserving transfer learning framework for kidney disease detection. Applied Sciences, 14(19), 8629.
  6. Chen, Z., Deng, J., Zhu, Q., Wang, H., & Chen, Y., 2022. A Systematic Review of Machine-Vision-Based Leather Surface Defect Inspection. Electronics, 11(15), 2383.
  7. Çam, K., Aydın, C., & Tarhan, C., 2022. Classification of Fabric Defects Using Deep Learning Algorithms. 2022 Innovations in Intelligent Systems and Applications Conference (ASYU), 1-6.
  8. Çeli̇k, H. İ., Dülger, L. C., Öztaş, B., Kertmen, M., & Gülteki̇N, E., 2022. A Novel Industrial Application of CNN Approach: Real Time Fabric Inspection and Defect Classification on Circular Knitting Machine. Tekstil ve Konfeksiyon, 32(4), 344-352.
  9. Ebayyeh, A. A. R. M. A., & Mousavi, A., 2020. A Review and Analysis of Automatic Optical Inspection and Quality Monitoring Methods in Electronics Industry. IEEE Access, 8, 183192-183271.
  10. Fang, B., Long, X., Sun, F., Liu, H., Zhang, S., & Fang, C., 2022. Tactile-Based Fabric Defect Detection Using Convolutional Neural Network with Attention Mechanism. IEEE Transactions on Instrumentation and Measurement, 71, 1-9.
  11. Geze, R. A., & Akbaş, A., 2024. Detection and Classification of Fabric Defects Using Deep Learning Algorithms. Politeknik Dergisi, 27(1), 371-378.
  12. Guder, O., Isik, S., & Anagun, Y., 2025. Fabric defects identification for textile industry with a deep learning approach. The Journal of The Textile Institute, 116(8), 1493-1502.
  13. Hanbay, K., 2023. Classification of Circular Knitting Fabric Defects Using MobileNetV2 Model. Türk Doğa ve Fen Dergisi, 12(4), 63-68.
  14. Hanbay, K., Talu, M. F., Özgüven, Ö. F., & Öztürk, D., 2019. Real-Time Detection of Knitting Fabric Defects Using Shearlet Transform. Tekstil Ve Konfeksiyon.
  15. Jia, Z., Shi, Z., Quan, Z., & Shunqi, M., 2022. Fabric defect detection based on transfer learning and improved Faster R-CNN. Journal of Engineered Fibers and Fabrics, 17, 15589250221086647.
  16. Jin, R., & Niu, Q., 2021. Automatic Fabric Defect Detection Based on an Improved YOLOv5. Mathematical Problems in Engineering, 2021(1),
  17. Khodier, M. M., Ahmed, S. M., & Sayed, M. S., 2022. Complex Pattern Jacquard Fabrics Defect Detection Using Convolutional Neural Networks and Multispectral Imaging. IEEE Access, 10, 10653-10660.
  18. Kuo, C. F. J., Wang, W. R., & Barman, J., 2022. Automated optical inspection for defect identification and classification in actual woven fabric production lines. Sensors, 22(19), 7246.
  19. Liu, Q., Wang, C., Li, Y., Gao, M., & Li, J., 2022. A Fabric Defect Detection Method Based on Deep Learning. IEEE access, 10, 4284-4296.
  20. Machado, R., Barros, L. A. M., Vieira, V., Silva, F. D. da, Costa, H., & Carvalho, V., 2025. Textile Defect Detection Using Artificial Intelligence and Computer Vision—A Preliminary Deep Learning Approach. Electronics, 14(18), 3692.
  21. Mehta, A., & Jain, R., 2023. An Analysis of Fabric Defect Detection Techniques for Textile Industry Quality Control. World Conference on Communication & Computing (WCONF), 1-5.
  22. Nasim, M., Mumtaz, R., Ahmad, M., & Ali, A., 2024. Fabric Defect Detection in Real World Manufacturing Using Deep Learning. Information, 15(8), 476.
  23. Nawar, S. H., Etawy, M. S., Nassar, G. E., Mohammed, N., & Hassabo, A. G., 2024. The Impact of CMF Design on Product Design. Journal of Textiles, Coloration and Polymer Science.
  24. Rasheed, A., Zafar, B., Rasheed, A., Ali, N., Sajid, M., Dar, S. H.,. & Mahmood, M. T., 2020. Fabric defect detection using computer vision techniques: a comprehensive review. Mathematical Problems in Engineering, 2020(1), 8189403.
  25. Saberironaghi, A., Ren, J., & El-Gindy, M., 2023. Defect Detection Methods for Industrial Products Using Deep Learning Techniques: A Review. Algorithms, 16(2), 95.
  26. Seeliger, A., Cheng, L., & Netland, T., 2023. Augmented reality for industrial quality inspection: An experiment assessing task performance and human factors. Computers in Industry, 151, 103985.
  27. Hussein, S. K., & Taifa, I. W., 2024. Six Sigma methodology implementation for minimising yarn breakages in the apparel industry. International Journal of Industrial and Systems Engineering.
  28. Soares, J. C., Sousa, S., & Tereso, A., 2020. Industry practices on the rework of defective products: Survey results. The TQM Journal, 32(6), 1177-1196.
  29. Wei, B., Hao, K., Tang, X., & Ding, Y., 2019. A new method using the convolutional neural network with compressive sensing for fabric defect classification based on small sample sizes. Textile Research Journal, 89(17), 3539-3555.
  30. Wen, Z., Zhao, Q., & Tong, L., 2020. CNN-based minor fabric defects detection. International Journal of Clothing Science and Technology, 33(1), 1-12.
  31. Zahra, A., Amin, M., El-Samie, F. E. A., & Emam, M., 2024. Efficient utilization of deep learning for the detection of fabric defects. Neural Computing and Applications, 36(11), 6037-6050.
  32. Zhang, J., Jing, J., Lu, P., & Song, S., 2022. Improved MobileNetV2-SSDLite for automatic fabric defect detection system based on cloud-edge computing. Measurement, 201, 111665.
  33. Rasheed, A., Zafar, B., Rasheed, A., Ali, N., Sajid, M., Dar, S. H.,. & Mahmood, M. T., 2020. Fabric defect detection using computer vision techniques: a comprehensive review. Mathematical Problems in Engineering, 2020(1), 8189403.