Topluluk Derin Öğrenme Mimarlarını Kullanarak Doğru Bitki Hastalığı Tanımlaması

Abstract

Bitki hastalıklarının erken teşhisi, mahsul sağlığını garanti altına almak ve tarımsal kayıpları azaltmak için hayati öneme sahiptir. Geleneksel görsel incelemenin manuel çabaya bağımlı olması, hem hızını hem de doğruluğunu doğal olarak sınırlandırmaktadır; bu da onu iyileştirme potansiyeli taşıyan bir alan haline getirmektedir. Bu zorluğun üstesinden gelmek için, bu çalışma, elma yaprağı hastalıklarının -sağlıklı, küllenme (powdery mildew) ve pas (rust) koşullarına odaklanarak- otomatik sınıflandırması amacıyla bir ensemble derin öğrenme çerçevesi sunmaktadır. Kamuya açık bir Kaggle veri setinden alınan 1.532 görüntü, çevirme, parlaklık ayarlama ve döndürme gibi teknikler kullanılarak 9.284 örneğe çıkarılmıştır. Beş önceden eğitilmiş evrişimli sinir ağı mimarisi—DenseNet201, EfficientNetB3, ResNet101, ResNet50 ve VGG16—doğruluk (accuracy), kesinlik (precision), duyarlılık (recall) ve F1-skoru metriklerine dayalı olarak ince ayar yapılarak değerlendirilmiştir. Bunlar arasında, EfficientNetB3 ve VGG16 tüm sınıflar boyunca üstün sınıflandırma performansı sergilemiştir. Önerilen ensemble modeli, %95.00 doğruluk oranına ve mükemmel kesinlik ve duyarlılık değerlerine (%100.00) ulaşarak, çeşitli temel ve ilgili çalışmaları geride bırakmıştır. Bu sonuçlar, hastalık tespit doğruluğunu artırmada ensemble derin öğrenme ve veri çoğaltma yöntemlerinin etkinliğini teyit etmekte olup, gerçek zamanlı bitki sağlığı izleme sistemleri için umut vaat eden bir çözüm sunmaktadır.

Keywords

• Görüntü işleme
• Bitki hastalığı
• Derin öğrenme modelleri
• Elma yaprağı sınıflandırması

Ethical Statement

Araştırmada kullanılan veri seti, Kaggle platformunda kamuya açık olarak sunulmaktadır ve kullanım koşullarına uygun şekilde değerlendirilmiştir. Bu nedenle etik kurul onayına gerek duyulmamıştır.

Comparative Analysis of CNN-Based Deep Learning Architectures for Automatic Plant Disease Classification

Abstract

Early detection of plant diseases is crucial for ensuring crop health and reducing agricultural losses. Traditional visual inspection presents a key opportunity for enhancement, as its dependence on manual effort naturally limits both its speed and accuracy. To address this challenge, this study conducts a comparative analysis of five convolutional neural network based architectures—DenseNet201, EfficientNetB3, ResNet101, ResNet50, and VGG16—for automatic classification of apple leaf diseases, focusing on healthy, powdery mildew, and rust conditions. A publicly available Kaggle dataset consisting of 1,532 images was augmented to 9,284 samples using techniques such as flipping, brightness adjustment, and rotation. Each model was fine-tuned and evaluated based on accuracy, precision, recall, and F1-score. Among these, EfficientNetB3 and VGG16 demonstrated superior classification performance across all classes, achieving up to 95.00% accuracy with perfect precision and recall (100.00%). These results confirm the effectiveness of transfer learning and data augmentation in enhancing disease detection accuracy, offering a promising foundation for real-time plant health monitoring systems.

Keywords

• Image processing
• Plant disease
• Deep learning models
• Apple leaf classification

Ethical Statement

The dataset used in this research is publicly available on Kaggle and was utilized in compliance with its terms of use. Therefore, ethical approval was not required.

References

1. [1] K. P. Ferentinos, “Deep learning models for plant disease detection and diagnosis,” Computers and Electronics in Agriculture, vol. 145, pp. 311–318, Feb. 2018, doi: 10.1016/j.compag.2018.01.009.
2. [2] A. A. Al-Zubi, “CNN-Based Detection of Powdery Mildew and Rust in Apple Orchards for Optimizing Crop Management,” Journal of Animal and Plant Sciences, vol. 35, no. 2, pp. 381–389, Apr. 2025, doi: 10.36899/JAPS.2025.2.0032.
3. [3] J. G. Arnal Barbedo, “Plant disease identification from individual lesions and spots using deep learning,” Biosystems Engineering, vol. 180, pp. 96–107, Apr. 2019, doi: 10.1016/j.biosystemseng.2019.02.002.
4. [4] S. Han Lee, H. Goëau, P. Bonnet, and A. Joly, “New perspectives on plant disease characterization based on deep learning,” Computers and Electronics in Agriculture, vol. 170, p. 105220, 2020, doi: 10.1016/j.compag.2020.105220.
5. [5] L. Li, S. Zhang, and B. Wang, “Plant Disease Detection and Classification by Deep Learning - A Review,” 2021, Institute of Electrical and Electronics Engineers Inc. doi: 10.1109/ACCESS.2021.3069646.
6. [6] S. ÖRENÇ, E. ACAR, and M. S. ÖZERDEM, “Utilizing the Ensemble of Deep Learning Approaches to Identify Monkeypox Disease,” DÜMF Mühendislik Dergisi, Jan. 2023, doi: 10.24012/dumf.1199679.
7. [7] G. Wang, Y. Sun, and J. Wang, “Automatic Image-Based Plant Disease Severity Estimation Using Deep Learning,” Computational Intelligence and Neuroscience, vol. 2017, 2017, doi: 10.1155/2017/2917536.
8. [8] S. Pudumalar and S. Muthuramalingam, “Hydra: An ensemble deep learning recognition model for plant diseases,” Journal of Engineering Research (Kuwait), vol. 12, no. 4, pp. 781–792, Dec. 2024, doi: 10.1016/j.jer.2023.09.033.

9. [9] S. Örenç, E. Acar, M. S. Özerdem, S. Şahin, and A. Kaya, “Automatic Identification of Adenoid Hypertrophy via Ensemble Deep Learning Models Employing X-ray Adenoid Images,” Journal of Imaging Informatics in Medicine, 2025, doi: 10.1007/s10278-025-01423-8.
10. [10] K. Nagasubramanian, S. Jones, A. K. Singh, S. Sarkar, A. Singh, and B. Ganapathysubramanian, “Plant disease identification using explainable 3D deep learning on hyperspectral images,” Plant Methods, vol. 15, no. 1, Aug. 2019, doi: 10.1186/s13007-019-0479-8.
11. [11] O. A. D. Ammar et al., “Plant Leaf Disease Detection Using Ensemble Learning and Explainable AI,” IEEE Access, vol. 12, pp. 156038–156049, 2024, doi: 10.1109/ACCESS.2024.3484574.
12. [12] F. Shahoveisi, H. Taheri Gorji, S. Shahabi, S. Hosseinirad, S. Markell, and F. Vasefi, “Application of image processing and transfer learning for the detection of rust disease,” Scientific Reports, vol. 13, no. 1, Dec. 2023, doi: 10.1038/s41598-023-31942-9.
13. [13] Ü. Atila, M. Uçar, K. Akyol, and E. Uçar, “Plant leaf disease classification using EfficientNet deep learning model,” Ecological Informatics, vol. 61, Mar. 2021, doi: 10.1016/j.ecoinf.2020.101182.
14. [14] T. Hayit, H. Erbay, F. Varçın, F. Hayit, and N. Akci, “Determination of the severity level of yellow rust disease in wheat by using convolutional neural networks,” Journal of Plant Pathology, vol. 103, no. 3, pp. 923–934, Aug. 2021, doi: 10.1007/s42161-021-00886-2.
15. [15] S. Sachar and A. Kumar, “Deep ensemble learning for automatic medicinal leaf identification,” International Journal of Information Technology (Singapore), vol. 14, no. 6, pp. 3089–3097, Oct. 2022, doi: 10.1007/s41870-022-01055-z.
16. [16] C. K. Sunil, C. D. Jaidhar, and N. Patil, “Systematic study on deep learning-based plant disease detection or classification,” Artificial Intelligence Review, vol. 56, no. 12, pp. 14955–15052, Dec. 2023, doi: 10.1007/s10462-023-10517-0.
17. [17] S. Orenc, M. S. Ozerdem, E. Acar, and M. Yilmaz, “Automatic segmentation of chest X-ray images via deep-improved various U-Net techniques,” Digital Health, vol. 11, May 2025, doi: 10.1177/20552076251366855.
18. [18] A. Sorayaie Azar, A. Naemi, S. Babaei Rikan, J. Bagherzadeh Mohasefi, H. Pirnejad, and U. K. Wiil, “Monkeypox detection using deep neural networks,” BMC Infectious Diseases, vol. 23, no. 1, Dec. 2023, doi: 10.1186/s12879-023-08408-4.
19. [19] S. Matarneh, F. Elghaish, F. Pour Rahimian, E. Abdellatef, and S. Abrishami, “Evaluation and optimisation of pre-trained CNN models for asphalt pavement crack detection and classification,” Automation in Construction, vol. 160, Apr. 2024, doi: 10.1016/j.autcon.2024.105297.
20. [20] A. Alshoraihy, “EfficientNetB3 in Leukemia Detection: Advancements in Medical Imaging Analysis,” Medinformatics, Feb. 2025, doi: 10.47852/bonviewmedin52023293.
21. [21] P. Utami, R. Hartanto, and I. Soesanti, “The EfficientNet Performance for Facial Expressions Recognition,” in 2022 5th International Seminar on Research of Information Technology and Intelligent Systems, ISRITI 2022, Institute of Electrical and Electronics Engineers Inc., 2022, pp. 756–762. doi: 10.1109/ISRITI56927.2022.10053007.
22. [22] K. Swetha and A. Seenu, “Deep Learning Approaches for Brain Tumor Detection in MRI Scans: A Comparative Study of DenseNet and ResNet101 Architectures,” in International Conference on Smart Systems for Applications in Electrical Sciences, ICSSES 2024, Institute of Electrical and Electronics Engineers Inc., 2024. doi: 10.1109/ICSSES62373.2024.10561400.
23. [23] M. Prabhakar, R. Purushothaman, and D. P. Awasthi, “Deep learning based assessment of disease severity for early blight in tomato crop,” Multimedia Tools and Application, vol. 79, no. 39–40, pp. 28773–28784, Oct. 2020, doi: 10.1007/s11042-020-09461-w.
24. [24] M. K. Türkeş and Y. Aydın, “Enhanced Emotion Recognition through Hybrid Deep Learning and SVM Integration,” Bitlis Eren Üniversitesi Fen Bilimleri Dergisi, vol. 14, no. 1, pp. 348–360, Mar. 2025, doi: 10.17798/bitlisfen.1588046.
25. [25] L. E. Demir and Y. Canbay, “Deepfake Image Detection with Transfer Learning Models,” Bitlis Eren Üniversitesi Fen Bilimleri Dergisi, vol. 14, no. 1, pp. 546–560, Mar. 2025, doi: 10.17798/bitlisfen.1610300.
26. [26] K. Simonyan and A. Zisserman, “Very Deep Convolutional Networks for Large-Scale Image Recognition,” Apr. 2015, [Online]. Available: http://arxiv.org/abs/1409.1556
27. [27] J. Sharma et al., “Deep learning based ensemble model for accurate tomato leaf disease classification by leveraging ResNet50 and MobileNetV2 architectures,” Scientific Reports, vol. 15, no. 1, Dec. 2025, doi: 10.1038/s41598-025-98015-x.