TÜRK KAHVESİNDE ŞEKER VE KAHVE SEVİYELERİNİN GÖRÜNTÜ TABANLI DERİN ÖZELLİKLER VE YÜZEY MORFOLOJİSİ ANALİZİYLE SINIFLANDIRILMASI

Abstract

Bu çalışmada, Türk kahvesinde şeker ve kahve miktarına bağlı formülasyon farklılıklarının görüntü tabanlı yöntemlerle belirlenebilirliği araştırılmıştır. Bu amaçla farklı şeker ve kahve seviyelerini temsil eden altı ayrı kahve formülasyonu hazırlanmış ve her gruba ait 500 adet görüntüden oluşan bir veri seti elde edilmiştir. İlk aşamada, görüntülerden önceden eğitilmiş ResNet-50 mimarisinin avg_pool katmanından derin özellikler çıkarılmış, bu özellikler ECOC tabanlı doğrusal destek vektör makineleri ile sınıflandırılmıştır. Model altı sınıflı formülasyon tanıma, üç sınıflı şeker seviyesi belirleme ve iki sınıflı kahve miktarı tahmini problemlerinde değerlendirilmiştir. Ayrıca özellik uzayındaki ayrışmayı incelemek amacıyla PCA ve t-SNE analizleri uygulanmıştır. İkinci aşamada ise sınıflandırma sonuçlarını desteklemek amacıyla kahve yüzeyine ait fraktal boyut, entropi, kontrast, homojenlik, köpük oranı ve parlaklık parametreleri hesaplanmıştır. Elde edilen bulgular, formülasyona bağlı değişimlerin derin özellik uzayında ayırt edilebilir örüntüler oluşturduğunu ve yüzey morfolojisi parametrelerinde sistematik değişimlerle ilişkili olduğunu göstermiştir.

Keywords

• Türk kahvesi
• derin özellikler
• ResNet-50
• destek vektör makineleri
• fraktal boyut
• görüntü analizi
• yüzey morfolojisi

CLASSIFICATION OF SUGAR AND COFFEE LEVELS IN TURKISH COFFEE USING IMAGE-BASED DEEP FEATURES AND SURFACE MORPHOLOGY ANALYSIS

Abstract

This study investigates the determinability of formulation differences in Turkish coffee depending on sugar and coffee content using image-based methods. For this purpose, six different coffee formulations representing different sugar and coffee levels were prepared, and a dataset consisting of 500 images for each group was obtained. In the first stage, deep features were extracted from the avg_pool layer of the pre-trained ResNet-50 architecture from the images, and these features were classified using ECOC-based linear support vector machines. The model was evaluated in six-class formulation recognition, three-class sugar level determination, and two-class coffee content estimation problems. In addition, PCA and t-SNE analyses were applied to examine the decomposition in the feature space. In the second stage, fractal dimension, entropy, contrast, homogeneity, foam ratio, and brightness parameters of the coffee surface were calculated to support the classification results. The findings indicated that formulation-dependent variations were associated with distinguishable patterns in the deep feature space and systematic changes in surface morphology parameters.

Keywords

• Turkish coffee
• deep features
• ResNet-50
• support vector machines
• fractal dimension
• image analysis
• surface morphology

References

1. Amanpour, A., Selli, S. (2016). Differentiation of Volatile Profiles and Odor Activity Values of Turkish Coffee and French Press Coffee. Journal of Food Processing and Preservation, 40(5), 1116-1124.
2. Angeloni, G., Guerrini, L., Masella, P., Bellumori., M, Daluiso., S, Parenti, A., Innocenti, M. (2019). What kind of coffee do you drink? An investigation on effects of eight different extraction methods. Food Research International, 116, 1327-1335.
3. Arzate-Vázquez, I., Chanona-Pérez, J. J., Calderón-Domínguez, G., Terres-Rojas, E., Garibay-Febles., V, Martínez-Rivas., A, Gutiérrez-López., G.F. (2012). Microstructural characterization of chitosan and alginate films by microscopy techniques and texture image analysis. Carbohydrate Polymers, 87(1), 289-299.
4. Clarke, Ronald., Vitzthum, O.G. (2008). Coffee : Recent Developments, 270. Geliş tarihi gönderen https://books.google. com/books/about/Coffee.html?hl=tr&id=jIFY_Pz8LH0C
5. Cordoba, N., Fernandez-Alduenda, M., Moreno, F. L., Ruiz, Y. (2020). Coffee extraction: A review of parameters and their influence on the physicochemical characteristics and flavour of coffee brews. Trends in Food Science & Technology, 96, 45-60.
6. Dadalı, C. (2021). Influence of Water Hardness on Volatile Compounds and Sensory Properties of Turkish Coffee. The Journal of Food, 46(5), 1183-1194.
7. De Vivo, A., Genovese, A., Tricarico, M.C., Aprea, A., Sacchi, R., Sarghini, F. (2022). Volatile compounds in espresso resulting from a refined selection of particle size of coffee powder. Journal of Food Composition and Analysis, 114, 104779.
8. Demir, Y., Bertan, S. (2023). Spatial distribution of Türkiye’s local Turkish coffee kinds. Journal of Ethnic Foods, 2023 10:1, 10(1), 32-.

9. Dilek, N.M., Babaoglu, A.S., Alp, H., Unal, K., Karakaya, M. (2025). Marination with espresso, filter, and Turkish coffee: A novel approach to improving the tenderness of spent hen meat. Poultry Science, 104(12), 105988.
10. Dowlati, M., Mohtasebi, S.S, Omid, M., Razavi, S.H, Jamzad, M., De La Guardia, M. (2013). Freshness assessment of gilthead sea bream (Sparus aurata) by machine vision based on gill and eye color changes. Journal of Food Engineering, 119(2), 277-287.
11. Hu, G., Peng, X., Gao, Y., Huang, Y., Li, X., Su, H., Qiu, M. (2020). Effect of roasting degree of coffee beans on sensory evaluation: Research from the perspective of major chemical ingredients. Food Chemistry, 331, 127329.
12. Kıvançlı, J., Elmacı, Y. (2016). Characterization of Turkish-Style Boiled Coffee Aroma by Gas Chromatography and Mass Spectrometry and Descriptive Analysis Techniques. International Journal of Food Properties, 19(8), 1671-1686.
13. Komes, D., Bušić, A. (2014). Antioxidants in Coffee. Processing and Impact on Antioxidants in Beverages, 25-32. Minz, P.S., Sawhney, I.K., Saini, C.S. (2020). Algorithm for processing high definition images for food colourimetry. Measurement, 158, 107670.
14. Nakilcioğlu-Taş, E. (2018). The Effects of Sugar Addition and Degree of Roast on the Bioactive Compounds and Antioxidant Activity of Turkish-Style Coffee Brews. Indian Journal of Pharmaceutical Education and Research, 52(3), 456-466.
15. Pereira, L. L., Guarçoni, R. C., da Luz, J. M. R., de Oliveira, A. C., Moreli, A. P., Filete, C. A., Paiva, G. de, Debona, D. G., Gomes, W. dos S., Cardoso, W. S., Berilli, S. da S., Oliveira, E. C. da S. (2023). Impacts of brewing methods on sensory perception and organoleptic compounds of coffee. Food Chemistry Advances, 2, 100185.
16. Pérez-Nieto, A., Chanona-Pérez, J. J., Farrera-Rebollo, R. R., Gutiérrez- López, G. F., Alamilla-Beltrán, L., Calderón-Domínguez., G. (2010). Image analysis of structural changes in dough during baking. LWT - Food Science and Technology, 43(3), 535-543.
17. Pua, A., Goh, R. M. V., Huang, Y., Tang, V. C. Y., Ee, K. H., Cornuz, M., Liu, S. Q., Lassabliere, B., Yu, B. (2022). Recent advances in analytical strategies for coffee volatile studies: Opportunities and challenges. Food Chemistry, 388, 132971.
18. Turan Ayseli, M., Kelebek, H., Selli, S. (2021). Elucidation of aroma-active compounds and chlorogenic acids of Turkish coffee brewed from medium and dark roasted Coffea arabica beans. Food Chemistry, 338, 127821.
19. Viciano-Tudela, S., Parra, L., Navarro-Garcia, P., Sendra, S., Lloret, J. (2023). Proposal of a New System for Essential Oil Classification Based on Low-Cost Gas Sensor and Machine Learning Techniques. Sensors 2023, Vol. 23, Page 5812, 23(13), 5812.
20. Wang, X., Lim, L. T. (2023). Effects of grind size, temperature, and brewing ratio on immersion cold brewed and French press hot brewed coffees. Applied Food Research, 3(2), 100334.
21. Wang, X., Lim, L. T., Tan, S., Fu, Y. (2019). Investigation of the factors that affect the volume and stability of espresso crema. Food Research International, 116, 668-675.
22. Won Kang, G., Piao, Z., Youn Ko, J. (2022). Effects of water types and roasting points on consumer liking and emotional responses toward coffee. Food Quality and Preference, 101, 104631.
23. Yang, X., Foegeding, E. A. (2010). Effects of sucrose on egg white protein and whey protein isolate foams: Factors determining properties of wet and dry foams (cakes). Food Hydrocolloids, 24(2-3), 227-238.
24. Yavuzer, E. (2021). Determination of fish quality parameters with low cost electronic nose. Food Bioscience, 41, 100948.
25. Yavuzer, E. (2023). Rapid detection of sea bass quality level with machine learning and electronic nose. International Journal of Food Science & Technology, 58(5), 2355-2359.
26. Yavuzer, E., Köse, M. (2022). Prediction of fish quality level with machine learning. International Journal of Food Science & Technology, 57(8), 5250-5255.
27. Yavuzer, E., Yaprak Uslu, D., Köse, M., Yetişen, M., Alaşalvar, H., Şimşek, H. İ. (2025). Determination of quality levels of fish oils recovered from trout waste using machine learning and odor sensors. Journal of Food Measurement and Characterization, 1-10.
28. Yüksel, A. N., Özkara Barut, K. T., Bayram, M. (2020). The effects of roasting, milling, brewing and storage processes on the physicochemical properties of Turkish coffee. LWT, 131(4), 109711.
29. Zhu, L., Spachos, P., Pensini, E., Plataniotis, K. N. (2021). Deep learning and machine vision for food processing: A survey. Current Research in Food Science, 4, 233-249.