Evrişimli sinir ağları ile mermer ve granit çeşitlerinin transfer öğrenme yöntemiyle sınıflandırılması.

Öz

Türkiye 120'den fazla farklı renk ve desende mermer rezervi ile dünya mermer potansiyelinin %40'na sahiptir. Ayrıca 20’ye yakın granit rezervleriyle de yüksek potansiyele sahip bir ülkedir. Mermer ve granit gibi taşların sınıflandırması benzer renk ve desen nedeniyle oldukça zordur. Bu durum sektör çalışanları ve müşterileri için büyük sorun teşkil etmektedir. Bu soruna çözüm getirmek için Türkiye’deki mermer ve granit çeşitlerini sınıflandıracak bir yöntem önerilmiştir. On ildeki taş ocakları ve satış mağazalarında dört cep telefonu ve iki fotoğraf makinesi ile veri seti oluşturulmuştur. Kullanılan 6 kamera ile lens ve görüntüleme teknoloji çeşitliliği sağlanmıştır. İdeal ve ideal olmayan desen-renk tonlarına sahip taş görüntüleri yanı sıra, yanlış depolama şartlarına maruz kalmış taş görüntüleri, ışık-gölge gibi fotoğraflama hataları da veri setine dahil edilmiştir. Çalışma sonucunda 88’i mermer, 14’ü granit olmak üzere toplamda 102 taşa ait 24480 görüntü içeren yeni bir veri seti oluşturulmuştur. Üç farklı veri artırma yöntemi ile görüntü sayısı 171360’a çıkarılmıştır. Veri setiyle 7 adet evrişimli sinir ağı transfer öğrenme yöntemi kullanılarak eğitilmiştir. Eğitimde ezberlemenin önüne geçebilmek ve ağların performansı ile güvenirliğini artırmak için Stratified Shuffle Split, K fold cross validation yöntemleri kullanılmıştır. En başarılı ağın sınıflandırma başarı oranı %97.4’tür.

Anahtar Kelimeler

• Mermer ve Granit Sınıflandırma
• Evrişimli Sinir Ağlar
• Önceden Eğitilmiş Ağlar
• Transfer Öğrenme

Classification Varieties of Marble and Granite by Convolutional Neural Networks with Transfer Learning Method

Öz

Turkey has 40% of the world's marble potential with its marble reserves in more than 120 different colors and designs. In addition, it is a country with a high potential with nearly 20 granite reserves. The classification of stones such as marble and granite is quite difficult due to the similar color and pattern. This situation poses a major problem for sector employees and customers. In order to solve this problem, a method has been proposed to classify marble and granite types in Turkey. Data sets with four mobile phones and two cameras have been created in quarries and retail stores in ten cities. Lens and imaging technology diversity was provided with 6 cameras used. In addition to stone images with ideal and non-ideal pattern-color tones, images of stones exposed to improper storage conditions and photographic errors such as light and shadow were also included in the data set. As a result of the study, a new data set containing 24480 images of 102 stones in total, 88 of which are marble and 14 of which are granite, was created. The number of images was increased to 171360 with three different data augmentation methods. With the data set, 7 convolutional neural networks were trained using transfer learning method. Stratified Shuffle Split, K fold cross validation methods were used to prevent memorization in training and to increase the performance and reliability of networks. The classification accuracy rate of the most successful network is 97.4%.

Anahtar Kelimeler

• Marble and Granite Classification
• Convolutional Neural Network
• Pre-Trained Networks
• Transfer Learning

Kaynakça

1. Arıkan M., Türkiye’de Mermercilik, Bilimsel Madencilik Dergisi, 2 (7), 463-468, 1962.
2. Delta Mermer. Geçmişten günümüze mermer. https://www.deltamarble.com/tr/gecmisten-gunumuze-mermer/. Yayın tarihi 2020. Erişim tarihi Kasım 29, 2020.
3. MTA Genel Müdürlüğü Fizibilite Etütleri Daire Başkanlığı. Madencilik Sektörüne Ait Temel Ekonomik Göstergeler. https://www.mta.gov.tr/v3.0/sayfalar/bilgi-merkezi/maden-serisi/img/TEG2019.pdf. Yayın tarihi 2019. Erişim tarihi Kasım 29, 2020.
4. İstanbul Maden İhracatçılar Birliği. 2020 Yılı Aylık Bazda Maden İhracat Verileri. https://www.imib.org.tr/tr/raporlar/ihracat-istatistikleri. Yayın tarihi 2020. Erişim tarihi Kasım 29, 2020.
5. Juan M.C.V.A., Balibrea T., Manuel L., Marble Slabs Quality Classification System using Texture Recognition and Neural Networks Methodology, European Symposium on Artificial Neural Networks, Bruges, 75-80, 1999.
6. Alajarín J. M., Delgado J. D. L., Balibrea L. M. T., Automatic system for quality-based classification of marble textures, IEEE Transactıons on Systems Man Cybernetıcs-Part C: Applications and Reviews, 35 (4), 488-497, 2005.
7. Selver M. A., Akay O., Ardalı E., Yavuz A. B., Önal O., Özden G., Cascaded and hierarchical neural networks for classifying surface images of marble slabs, IEEE Transactions on Systems, Man, and Cybernetics, Part C (Applications and Reviews), 39 (4), 426-439, 2009.
8. Topalova I.C., Tzokev A., Adaptive marble plate classification system based on neural network and PLC Implementation, Proceedings of the 22nd International DAAAM Symposium ''Intelligent Manufacturing & Automation'', Vienna, 22 (1), 453-454, 2011.

9. Ferreira A., Girald G., Convolutional Neural Network approaches to granite tiles classification, Expert Systems With Applications, 84, 1-11. 2017.
10. Kaynar O., Torun Y., Temiz M., Görmez Y., Automatic Classification of Natural Stone Tiles with Computer Vision, 3rd International Conference on Computer Science and Engineering (UBMK), Saraybosna, 527–532, 2018.
11. Torun Y., Akbaş M.R., Çelik M.A., Kaynar O., Mermer Sınıflandırılması İçin Makineli Görü Sistem Geliştirilmesi, 2019 27th Signal Processing and Communications Applications Conference (SIU), Sivas, 5112520, 2019.
12. Pençe İ., Çeşmeli M. Ş., Deep Learning in Marble Slabs Classification, Techno-Science, 2 (1), 21-26, 2019.
13. Ather M., Khan B., Wang Z., Song G., Automatic Recognition and Classification of Granite Tiles Using Convolutional Neural Networks (CNN), ICAAI 2019: Proceedings of the 2019 3rd International Conference on Advances in Artificial Intelligence, Northumbria University, İstanbul, 193–197, 2019.
14. Karaali İ., Eminağaoğlu M., A convolutional neural network model for marble quality classification, Journal of the Faculty of Engineering and Architecture of Gazi University ,36 (1), 347-357, 2021.
15. Canayaz M., Uludağ F., Marble Classification Using Deep Neural Networks, European Journal of Technique (EJT), 10 (1), 52-63, 2020.
16. Arı A., Hanbay D., Tumor detection in MR images of regional convolutional neural networks, Journal of the Faculty of Engineering and Architecture of Gazi University, 34 (3), 1395-1408, 2019.
17. Dandıl E., Polattimur R., Dog Behavior Recognition and Tracking based on Faster R-CNN, Journal of the Faculty of Engineering and Architecture of Gazi University, 35 (2), 819-834, 2020.
18. Yıldız O., Melanoma detection from dermoscopy images with deep learning methods: A comprehensive study, Journal of the Faculty of Engineering and Architecture of Gazi University, 34 (4), 2241-2260, 2019.
19. Bingöl K., Akan A.E., Örmecioğlu H.T., Er A., Artificial intelligence applications in earthquake resistant architectural design: Determination of irregular structural systems with deep learning and ImageAI method, Journal of the Faculty of Engineering and Architecture of Gazi University, 35 (4), 2197-2209, 2020.
20. Aktaş A., Doğan B., Demir Ö., Tactile paving surface detection with deep learning methods, Journal of the Faculty of Engineering and Architecture of Gazi University, 35 (3), 1685-1700, 2020.
21. Lawrence S., Giles C.L., Tsoi A.C., Back A.D., Face recognition: A convolutional neural-network approach, IEEE Transactions on Neural Networks, 8 (1), 98-113, 1997.
22. MathWorks. Introduction to convolutional neural networks. https://www.mathworks.com/help/deeplearning/ug/introduction-to-convolutional-neural-networks.html. Yayın tarihi 2019. Erişim tarihi Kasım 26, 2019.
23. Baykal E., Dogan H., Ercin M.E., Ersoz S., Ekinci M., Transfer learning with pre-trained deep convolutional neural networks for serous cell classification, Multimedia Tools and Applications, 1-19, 2019.
24. Nebauer, C., Evaluation of convolutional neural networks for visual recognition, IEEE Transactions on Neural Networks, 9 (4), 685- 696, 1998.
25. Zhou Y., Nejati H., Do T.T., Cheung N.M., Cheah L., Image-based vehicle analysis using deep neural network: A systematic study, IEEE international conference on digital signal processing, Beijing, 276-280. 2016.
26. He K., Zhang X., Ren S., Sun J., Deep Residual Learning for Image Recognition, IEEE Conference on Computer Vision and Pattern Recognition, IEEE Computer Society, Las Vegas, 770-778, 2016.
27. Ye H., Han H., Zhu L., Duan Q., Vegetable pest image recognition method based on improved VGG convolution neural network, Journal of Physics: Conference Series, Xi’an, 1237 (3), 2019.
28. Theckedath D., Sedamkar R.R., Detecting Affect States Using VGG16, ResNet50 and SE‑ResNet50 Networks, Springer Nature Computer Science, 79, 1-7, 2020.
29. Pan S.J., Yang Q., A survey on transfer learning, IEEE Transactions on Knowledge and Data Engineering, 22 (10), 1345-1359, 2010.
30. Chollet F. Deep Learning with Python, Manning Publications, New York, 2017.
31. Kaggle. Stratified Shuffle Split Working with less data. https://www.kaggle.com/viswanathanc/stratifiedshufflesplit-working-with-less-data. Yayın tarihi 2019. Erişim tarihi Ekim 12, 2020.
32. Medium, StratifiedKFold v.s KFold v.s, https://xzz201920.medium.com/stratifiedkfold-v-s-kfold-v-s-stratifiedshufflesplit-ffcae5bfdf. Erişim tarihi Ekim 12, 2020.
33. Github, Cross-Validation: Why and how to do it, https://srikarvaka.github.io/model-evaluation/Cross-validation/. Yayın tarihi Ocak 29, 2019. Erişim tarihi Ekim 12, 2020
34. Mash R., Borghetti B., Pecarina J., Improved Aircraft Recognition for Aerial Refueling through Data Augmentation in Convolutional Neural Networks, 12th International Symposium on Visual Computing Advances in Visual Computing, Las Vegas, 12, 113-122, 2016.