Zeytin Olgunluk Seviyelerinin Derin Öğrenme ile Tespiti ve Yeni Bir Veri Kümesi

Abstract

Zeytin meyvesinin olgunluk seviyesi, zeytinyağı kalitesini doğrudan etkileyen en kritik parametrelerden biridir. Hasat sonrası zeytinlerin doğru ve tutarlı şekilde sınıflandırılması, üretim sürecinde kalite kontrolün sağlanması açısından büyük önem taşımaktadır. Ancak geleneksel yöntemler çoğunlukla operatör deneyimine dayalı olup öznel ve tutarsız sonuçlar üretebilmektedir. Bu çalışmada, zeytin meyvelerinin Yeşil, Ala Yeşil ve Siyah olmak üzere üç farklı olgunluk seviyesine göre otomatik olarak sınıflandırılmasına yönelik derin öğrenme tabanlı bir yaklaşım önerilmiştir. Çalışma kapsamında saha koşullarında elde edilen toplam 656 görüntüden oluşan özgün bir veri seti oluşturulmuştur. Veri seti üzerinde DenseNet121, InceptionV3, VGG16, VGG19, ResNet50 ve ResNet101 olmak üzere farklı transfer öğrenme mimarileri kullanılarak karşılaştırmalı analiz gerçekleştirilmiştir. Elde edilen deneysel sonuçlar, DenseNet121 modelinin %98,48 doğruluk, %98,53 kesinlik, %98,48 duyarlılık ve %98,47 F1-skoru ile en yüksek performansı sağladığını göstermektedir. Sonuçlar, derin öğrenme tabanlı yöntemlerin zeytin olgunluk seviyelerinin otomatik belirlenmesinde yüksek doğruluk ve güvenilirlik sunduğunu ortaya koymaktadır

Keywords

• Zeytin olgunluğu
• Derin öğrenme
• Transfer öğrenme
• Tarımsal kalite kontrol

Detection of Olive Ripeness Levels Using Deep Learning and a New Dataset

Abstract

The ripeness level of the olive fruit is one of the most critical parameters directly affecting olive oil quality. The accurate and consistent classification of olives after harvest is of great importance for ensuring quality control during the production process. However, traditional methods are largely based on operator experience and can produce subjective and inconsistent results. In this study, a transfer learning-based approach is proposed for the automatic classification of olive fruits into three different ripeness levels: Green, Light Green, and Black. A unique dataset consisting of a total of 656 images obtained under field conditions was created for this study. A comparative analysis was conducted using different deep learning architectures—including DenseNet121, InceptionV3, VGG16, VGG19, ResNet50, and ResNet101—on the dataset. The experimental results indicate that the DenseNet121 model achieved the highest performance with 98.48% accuracy, 98.53% precision, 98.48% sensitivity, and an F1-score of 98.47. The results demonstrate that deep learning-based methods offer high accuracy and reliability in the automatic determination of olive ripeness levels.

Keywords

• Olive ripeness
• Deep learning
• Transfer learning
• Agricultural quality control

References

1. [1] J. M. Hernández-Mogollón, E. Di-Clemente, A. M. Campón-Cerro, and J. A. Folgado-Fernández, “Olive oil tourism in the Euro-Mediterranean area,” International Journal of Euro-Mediterranean Studies, vol. 14, no. 1, 2021.
2. [2] J. Gagour, O. Hallouch, A. Asbbane, L. Bijla, A. Laknifli, L. H. Lee and S. Gharby, “A review of recent progresses on olive oil chemical profiling, extraction technology, shelf-life, and quality control,” Chemistry & Biodiversity, vol. 21, no. 4, Art. no. e202301697, 2024.
3. [3] T. Lechhab, W. Lechhab, F. Cacciola, and F. Salmoun, “Sets of internal and external factors influencing olive oil (Olea europaea L.) composition: A review,” European Food Research and Technology, vol. 248, no. 4, pp. 1069–1088, 2022.
4. [4] C. Skodra, V. S. Titeli, M. Michailidis, C. Bazakos, I. Ganopoulos, A. Molassiotis, and G. Tanou, “Olive fruit development and ripening: Break on through to the ‘-omics’ side,” International Journal of Molecular Sciences, vol. 22, no. 11, p. 5806, 2021.
5. [5] M. G. Emmanouilidou, M. Koukourikou-Petridou, D. Gerasopoulos, and M. C. Kyriacou, “Oxidative stability, fatty-acid and phenolic composition of Cypriot monovarietal virgin olive oils with progressive fruit maturity,” Journal of Food Composition and Analysis, vol. 104, p. 104191, 2021.
6. [6] E. Yilmaz and A. Aydin, “Production of early and normal harvest olive oils under industrial conditions and comprehensive comparison of the oils produced,” European Journal of Lipid Science and Technology, vol. 127, no. 9, Art. no. e70052, 2025.
7. [7] L. Ortenzi, S. Figorilli, C. Costa, F. Pallottino, S. Violino, M. Pagano and F. Antonucci, “A machine vision rapid method to determine the ripeness degree of olive lots,” Sensors, vol. 21, no. 9, p. 2940, 2021.
8. [8] L. Jiang, B. Xu, N. Husnain, and Q. Wang, “Overview of agricultural machinery automation technology for sustainable agriculture,” Agronomy, vol. 15, no. 6, p. 1471, 2025.

9. [9] R. Archana and P. E. Jeevaraj, “Deep learning models for digital image processing: A review,” Artificial Intelligence Review, vol. 57, no. 1, p. 11, 2024.
10. [10] C. Wang, S. Liu, Y. Wang, J. Xiong, Z. Zhang, B. Zhao and P. He, “Application of convolutional neural network-based detection methods in fresh fruit production: A comprehensive review,” Frontiers in Plant Science, vol. 13, p. 868745, 2022.
11. [11] K. Weiss, T. M. Khoshgoftaar, and D. Wang, “A survey of transfer learning,” Journal of Big Data, vol. 3, no. 1, p. 9, 2016.
12. [12] S. Rad, M. Jalili, F. Ansari, R. Nodeh, and L. Rashidi, “Maturity impact on physicochemical composition and polyphenol properties of extra virgin olive oils,” Food Science & Nutrition, vol. 11, pp. 5396–5408, 2023.
13. [13] M. Emmanouilidou, M. Koukourikou-Petridou, D. Gerasopoulos, and M. Kyriacou, “Evolution of physicochemical constitution and cultivar-differential maturity configuration in olive fruit,” Scientia Horticulturae, vol. 272, p. 109516, 2020.
14. [14] J. Ezenarro, Á. García-Pizarro, O. Busto, A. De Juan, and R. Boqué, “Analysing olive ripening with digital image RGB histograms,” Analytica Chimica Acta, vol. 1280, p. 341884, 2023.
15. [15] G. Salvucci, F. Pallottino, L. De Laurentiis, F. Del Frate, R. Manganiello, F. Tocci, S. Vasta, S. Figorilli, B. Bassotti, S. Violino, L. Ortenzi and F. Antonucci, “Fast olive quality assessment through RGB images and advanced convolutional neural network modeling,” European Food Research and Technology, vol. 248, pp. 1395–1405, 2022.
16. [16] E. Kutlu and F. Şen, “Farklı hasat zamanlarının Gemlik zeytin çeşidinde kaliteye etkileri,” Ege Üniversitesi Ziraat Fakültesi Dergisi, vol. 48, no. 2, pp. 85–93, 2011.
17. [17] N. B. Youssef, W. Zarrouk, A. Carrasco-Pancorbo, Y. Ouni, A. Segura-Carretero, A. Fernández-Gutiérrez, D. Daoud and M. Zarrouk, “Effect of olive ripeness on chemical properties and phenolic composition of Chetoui virgin olive oil,” Journal of the Science of Food and Agriculture, vol. 90, no. 2, pp. 199–204, 2010.
18. [18] E. B. Uyanik and A. Gümüşkesen, “Erkence zeytin çeşidinde olgunluk düzeyinin etkisi,” Türk Tarım ve Doğa Bilimleri Dergisi, vol. 11, no. 1, pp. 180–189, 2024.
19. [19] D. Bozdoğan Konuşkan, “Zeytin olgunluğunun kalite parametreleri üzerine etkisi,” Turkish Journal of Agriculture - Food Science and Technology, vol. 9, pp. 2493–2497, 2022.
20. [20] A. Acar, “Zeytinyağı eldesinde üretim koşulları ve kalite etkileri,” Ph.D. dissertation, Necmettin Erbakan Univ., 2023.
21. [21] E. Doğru, Ş. Çelik, Y. Yakar and N. Ünver, “Zeytin yaprağı ilavesinin zeytinyağı özelliklerine etkisi,” Harran Tarım ve Gıda Bilimleri Dergisi, vol. 25, no. 1, pp. 72–85, 2021.
22. [22] M. Criado, M. Motilva, M. Goñi, and M. Romero, “Effect of olive fruit maturation on pigment fractions,” Food Chemistry, vol. 100, pp. 748–755, 2005.
23. [23] S. Ali, M. Mohamed, F. Bousorra, S. Tlahig, T. Triki, N. Kamoun, A. Lazzez, F. Guasmi and K. Nagaz, “Physicochemical profiling of olive oil as affected by variety and maturity,” Polish Journal of Environmental Studies, 2025.
24. [24] L. Ortenzi, S. Figorilli, C. Costa, F. Pallottino, S. Violino, M. Pagano, F. Antonucci, “A machine vision rapid method to determine ripeness,” Sensors, vol. 21, no. 9, 2021.
25. [25] A. Beyaz and R. Öztürk, “Identification of olive cultivars using image processing techniques,” Turkish Journal of Agriculture and Forestry, vol. 40, pp. 671–683, 2016.
26. [26] F. B. K. İnce, S. Taşdemir, and İ. A. Özkan, “Classification of olive fruit with image processing techniques,” Selçuk University Journal of Engineering Sciences, vol. 19, no. 4, pp. 156–167, 2020.
27. [27] S. Gönenç and Y. Öner, “Zeytin türlerinin belirlenmesinde görüntü işleme,” The Journal of International Scientific Researches, vol. 6, no. 2, pp. 137–151, 2021.
28. [28] J. Ezenarro, Á. García-Pizarro, O. Busto, A. de Juan and R. Boqué, “Analysing olive ripening with RGB histograms,” Analytica Chimica Acta, vol. 1280, p. 341884, 2023.
29. [29] S. Figorilli, S. Violino, L. Moscovini, L. Ortenzi, G. Salvucci, S. Vasta, F. Tocci, C. Costa, P. Toscano and F. Pallottino, “Olive fruit selection through AI algorithms,” Foods, vol. 11, no. 21, 2022.
30. [30] L. Gargiulo, A. G. Napolitano, S. M. Marsala, A. Di Salle and G. Mele, “Olive variety classification from 3D morphology,” Food Science & Nutrition, vol. 13, no. 9, Art. no. e70797, 2025.
31. [31] H. Miho, G. Pagnotta, D. Hitaj, F. De Gaspari, L. V. Mancini, G. Koubouris, G. Godino, M. Hakan and C. M. Diez, “OliVaR: Improving olive variety recognition using deep neural networks,” Computers and Electronics in Agriculture, vol. 216, p. 108530, 2024.
32. [32] B. Dikici, M. F. Bekçioğulları, H. Açıkgöz and D. Korkmaz, “Zeytin yaprağı hastalık sınıflandırması,” Konya Journal of Engineering Sciences, vol. 10, no. 3, pp. 535–547, 2022.
33. [33] S. Dulkadir and G. K. Gültekin, “Muz olgunluk seviyelerinin derin öğrenme ile sınıflandırılması,” EMO Bilimsel Dergi, vol. 13, no. 3, pp. 27–34, 2023.
34. [34] K. Bozoğlan and U. Yüzgeç, “Görüntü sınıflandırmada performans karşılaştırması,” Mühendislik Bilimleri ve Tasarım Dergisi, vol. 13, no. 3, pp. 856–871, 2025.
35. [35] K. Sabancı and C. Aydın, “Using image processing and ANN to classify olives,” Tarım Makinaları Bilimi Dergisi, vol. 10, no. 3, pp. 243–246, 2014.
36. [36] A. Raza, M. S. Alshehri, S. Almakdi, A. A. Siddique, M. Alsulami and M. Alhaisoni, “Enhancing brain tumor classification with transfer learning,” International Journal of Imaging Systems and Technology, vol. 34, no. 1, Art. no. e22957, 2024.
37. [37] I. Khalil, A. Mehmood, H. Kim and J. Kim, “OCTNet: Multi-scale attention network,” Mathematics, vol. 12, no. 19, p. 3003, 2024.
38. [38] L. Kong and J. Cheng, “COVID-19 X-ray classification using DenseNet,” Biomedical Signal Processing and Control, vol. 77, p. 103772, 2022.
39. [39] Q. Zhang, “A novel ResNet101 model for image classification,” SN Applied Sciences, vol. 4, no. 1, p. 9, 2022.
40. [40] S. Safaee, Z. Libohova, E. J. Kladivko, A. Brown, E. Winzeler, Q. Read, K. Veum and K. Adhikari, “Prediction accuracy of soil properties,” Geoderma Regional, vol. 36, Art. no. e00766, 2024.
41. [41] T. Silva Filho, H. Song, M. Perello-Nieto, R. Santos-Rodriguez, M. Kull and P. A. Flach, “Classifier calibration: A survey,” Machine Learning, vol. 112, no. 9, pp. 3211–3260, 2023.
42. [42] G. M. Foody, “Challenges in classification accuracy metrics,” PLOS ONE, vol. 18, no. 10, Art. no. e0291908, 2023.
43. [43] J. C. Obi, “Comparative study of classification metrics,” World Journal of Advanced Engineering Technology and Sciences, vol. 8, no. 1, pp. 308–314, 2023.
44. [44] Y. Canbay, S. Adsiz, and P. Canbay, “Privacy-preserving transfer learning framework,” Applied Sciences, vol. 14, no. 19, p. 8629, 2024.