A Hybrid Framework Based on Transfer Learning for Seasonal Classification of Satellite Images

Abstract

Seasonal classification from satellite imagery is a crucial remote sensing task for monitoring vegetation dynamics, agricultural processes, environmental change, and climate-related spatial patterns. However, limited labeled data, regional variability, and the high computational cost of training large-scale networks from scratch make developing generalizable models for this task challenging. This study proposes a hybrid framework based on transfer learning for seasonal classification using monthly satellite imagery from all 81 provinces of Türkiye. Eight ImageNet pretrained deep learning models were evaluated as feature extractors, and the extracted representations were evaluated using seven machine learning classifiers. The selection of feature extractors and classifiers was performed using 10-fold cross-validation repeated five times. Among the feature extractor models, the combination of ConvNeXt and Multi-Layer Perceptron produced the best result with a mean accuracy of 0.698 and a mean F1-score of 0.696. As a result of the comparative analysis, ConvNeXt, Vision Transformer, and Swin Transformer were selected as the three most successful feature extractors, and Multi-Layer Perceptron was selected as the final classifier. The hybrid model, constructed by concatenating the features of these three selected feature extractors, achieved a mean accuracy of 0.725 and a mean F1-score of 0.724. After fine-tuning, the proposed hybrid model achieved the highest performance with a mean accuracy of 0.740 and a mean F1-score of 0.742. The results demonstrate that combining pretrained representations and fine-tuning improved seasonal classification in satellite imagery. The proposed framework offers an effective and computationally practical approach for seasonal classification, providing a promising foundation for future environmental monitoring and agricultural remote sensing applications. 

Keywords

• Transfer learning
• Pretrained models
• Machine learning classification
• satellite image classification

Uydu Görüntülerinin Mevsimsel Sınıflandırılması için Transfer Öğrenmeye Dayalı Hibrit Çerçeve

Abstract

Uydu görüntülerinden mevsimsel sınıflandırma, bitki örtüsü dinamiklerini, tarımsal süreçleri, çevresel değişimi ve iklimle ilgili mekânsal desenleri izlemek için önemli bir uzaktan algılama görevidir. Bununla birlikte, sınırlı sayıda etiketli veri, bölgesel değişkenlik ve büyük ölçekli ağların sıfırdan eğitilmesinin yüksek hesaplama maliyeti, bu görev için genelleştirilebilir modeller oluşturmayı zorlaştırmaktadır. Bu çalışma, Türkiye'nin 81 ilinin tamamından elde edilen aylık uydu görüntülerini kullanarak mevsimsel sınıflandırma için transfer öğrenmeye dayalı hibrit bir çerçeve önermektedir. Sekiz adet ImageNet ön eğitilmiş derin öğrenme modeli özellik çıkarıcı olarak değerlendirilmiş ve çıkarılan temsiller yedi makine öğrenmesi sınıflandırıcısı kullanılarak sınıflandırılmıştır. Özellik çıkarıcıların ve sınıflandırıcıların seçimi, beş kez tekrarlanan 10 katlı çapraz doğrulama kullanılarak gerçekleştirilmiştir. Özellik çıkarıcı modeller arasında, ConvNeXt ve çok katmanlı algılayıcı kombinasyonu, ortalama 0,698 doğruluk ve ortalama 0,696 F1 puanı ile en iyi sonucu vermiştir. Karşılaştırmalı analiz sonucunda, ConvNeXt, Vision Transformer ve Swin Transformer en başarılı üç özellik çıkarıcı olarak seçilmiş ve çok katmanlı algılayıcı nihai sınıflandırıcı olarak belirlenmiştir. Seçilen bu üç özellik çıkarıcının özelliklerinin birleştirilmesiyle oluşturulan hibrit model, ortalama 0,725 doğruluk ve ortalama 0,724 F1 puanı elde etmiştir. İnce ayardan sonra, önerilen hibrit model, ortalama 0,740 doğruluk ve ortalama 0,742 F1 puanı ile en yüksek performansı göstermiştir. Sonuçlar, önceden eğitilmiş temsillerin birleştirilmesi ve ince ayarın, uydu görüntülerinde mevsimsel sınıflandırmayı iyileştirdiğini göstermektedir. Önerilen çerçeve, mevsimsel sınıflandırma için etkili ve hesaplama açısından pratik bir yaklaşım sunarak, gelecekteki çevresel izleme ve tarımsal uzaktan algılama uygulamaları için umut vadeden bir temel sağlamaktadır. 

Keywords

• Transfer öğrenme
• Ön eğitilmiş modeller
• Makine öğrenmesi sınıflandırması
• Uydu görüntüsü sınıflandırması

References

1. [1] Krejcar, O., Namazi, H. (2026). AI in remote sensing and satellite image processing-a review. Environmental Earth Sciences. 85(3). 78. https://doi.org/10.1007/s12665-025-12798-w.
2. [2] LeCun, Y., Bengio, Y., Hinton, G. (2015). Deep learning. Nature. 521(7553), 436-444. https://doi.org/10.1038/nature14539.
3. [3] Zhu, X. X., Tuia, D., Mou, L., Xia, G. S., Zhang, L., Xu, F., Fraundorfer, F. (2017). Deep learning in remote sensing: A comprehensive review and list of resources. IEEE geoscience and remote sensing magazine. 5(4). 8-36. 10.1109/MGRS.2017.2762307.
4. [4] Yuan, Q., Shen, H., Li, T., Li, Z., Li, S., Jiang, Y., Xu, H., Tan, W., Yang, Q., Wang, J., Gao, J. (2020). Deep learning in environmental remote sensing: Achievements and challenges. Remote sensing of Environment. 241. 111716. 10.1016/j.rse.2020.111716.
5. [5] Zhao, T., Wang, S., Ouyang, C., Chen, M., Liu, C., Zhang, J., Yu, L., Wang, F., Xie, Y., Li, J., Wang, F. (2024). Artificial intelligence for geoscience: Progress, challenges, and perspectives. The Innovation. 5(5). 10.1016/j.xinn.2024.100691.
6. [6] Cheng, G., Han, J., Lu, X. (2017). Remote sensing image scene classification: Benchmark and state of the art. Proceedings of the IEEE. 105(10). 1865-1883. 10.1109/JPROC.2017.2675998.
7. [7] Moharram, M. A., Sundaram, D. M. (2023). Land use and land cover classification with hyperspectral data: A comprehensive review of methods, challenges and future directions. Neurocomputing. 536. 90-113. https://doi.org/10.1016/j.neucom.2023.03.025.
8. [8] Gui, S., Song, S., Qin, R., Tang, Y. (2024). Remote sensing object detection in the deep learning era - a review. Remote Sensing. 16(2). 327. https://doi.org/10.3390/rs16020327.

9. [9] Li, K., Wan, G., Cheng, G., Meng, L., Han, J. (2020). Object detection in optical remote sensing images: A survey and a new benchmark. ISPRS journal of photogrammetry and remote sensing. 159. 296-307. https://doi.org/10.1016/j.isprsjprs.2019.11.023.
10. [10] Uzun, S., Arslantaş, H. (2024). Determination of Radiation Value by Month Using Artificial Neural Network Model; Ankara, Sivas, Erzurum example. Gazi Üniversitesi Fen Bilimleri Dergisi Part C: Tasarım ve Teknoloji. 12(1). 315-323. 10.29109/gujsc.1420617.
11. [11] Pires de Lima, R., Marfurt, K. (2019). Convolutional neural network for remote-sensing scene classification: Transfer learning analysis. Remote Sensing. 12(1). 86. https://doi.org/10.3390/rs12010086.
12. [12] Thirumaladevi, S., Swamy, K. V., Sailaja, M. (2023). Remote sensing image scene classification by transfer learning to augment the accuracy. Measurement: Sensors. 25. 100645. https://doi.org/10.1016/j.measen.2022.100645.
13. [13] Kunwar, S., Ferdush, J. (2024). Mapping of Land Use and Land Cover (LULC) Using EuroSAT and Transfer Learning. Revue Internationale de Geomatique. 33. 1-13. https://doi.org/10.32604/rig.2023.047627.
14. [14] Roy, A., Gourisaria, M. K., Chatterjee, R., Jha, A. V., Appasani, B., Bizon, N., Mazare, A. G. (2025). A privacy-preserving, on-board satellite image classification technique incorporating homomorphic encryption and transfer learning. Frontiers in Remote Sensing. 6. 1678882. https://doi.org/10.3389/frsen.2025.1678882.
15. [15] Farag, N. H., Abdelhafiz, D., Abdrabo, M. A., ELIskandarani, M. A., Hassaan, M. A. (2026). Enhancing the classification of spectrally similar land use/land cover classes using transfer learning in arid regions. Scientific Reports. https://doi.org/10.1038/s41598-026-38540-5.
16. [16] Sudiana, D., Lestari, A. I., Riyanto, I., Rizkinia, M., Arief, R., Prabuwono, A. S., Sri Sumantyo, J. T. (2023). A hybrid convolutional neural network and random forest for burned area identification with optical and synthetic aperture radar (SAR) data. Remote Sensing. 15(3). 728. https://doi.org/10.3390/rs15030728.
17. [17] Kwenda, C., Gwetu, M., Fonou-Dombeu, J. V. (2024). Hybridizing deep neural networks and machine learning models for aerial satellite forest image segmentation. Journal of imaging. 10(6). 132. https://doi.org/10.3390/jimaging10060132.
18. [18] Mashraqi, A. M., Alshari, E. A., Halawani, H. T., Senan, E. M., Asiri, Y., Alowadhi, B. M. (2025). Hybrid deep learning and optimization-based land use and land cover classification for advancing sustainable agriculture in Najran city, Saudi Arabia. Scientific Reports. 15(1). 41808. https://doi.org/10.1038/s41598-025-25908-2.
19. [19] Sudiana, D., Putri, S. H., Kushardono, D., Prabuwono, A. S., Sri Sumantyo, J. T., Rizkinia, M. (2025). CNN-random forest hybrid method for phenology-based paddy rice mapping using Sentinel-2 and Landsat-8 satellite images. Computers. 14(8). 336. https://doi.org/10.3390/computers14080336.
20. [20] Cheng, P., Zhou, J. (2011). Automatic season classification of outdoor photos. Third International Conference on Intelligent Human-Machine Systems and Cybernetics. 1. 46-49. IEEE.
21. [21] Volokitin, A., Timofte, R., Van Gool, L. (2016). Deep features or not: Temperature and time prediction in outdoor scenes. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition Workshops. 63-71.
22. [22] Bilotta, G., Bibbò, L., Meduri, G. M., Genovese, E., Barrile, V. (2025). Deep learning innovations: Resnet applied to SAR and sentinel-2 imagery. Remote Sensing. 17(12). 1961. https://doi.org/10.3390/rs17121961.
23. [23] He, K., Zhang, X., Ren, S., Sun, J. (2016). Deep residual learning for image recognition. In Proceedings of the IEEE conference on computer vision and pattern recognition. 770-778.
24. [24] Huang, G., Liu, Z., Van Der Maaten, L., Weinberger, K. Q. (2017). Densely connected convolutional networks. In Proceedings of the IEEE conference on computer vision and pattern recognition. 4700-4708.
25. [25] Kurtoğlu, B., Çiftçi, S. (2024). A comparative study of swin-based enhanced remote sensing image classifications. In Proceedings of the 2024 6th International Conference on Image Processing and Machine Vision. 68-74.
26. [26] Liu, Z., Mao, H., Wu, C. Y., Feichtenhofer, C., Darrell, T., Xie, S. (2022). A convnet for the 2020s. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. 11976-11986.
27. [27] Liu, H. I., Galindo, M., Xie, H., Wong, L. K., Shuai, H. H., Li, Y. H., Cheng, W. H. (2024). Lightweight deep learning for resource-constrained environments: A survey. ACM Computing Surveys. 56(10). 1-42. https://doi.org/10.1145/3657282.
28. [28] Howard, A., Sandler, M., Chu, G., Chen, L.C., Chen, B., Tan, M., Wang, W., Zhu, Y., Pang, R., Vasudevan, V., Le, Q.V. (2019). Searching for mobilenetv3. In Proceedings of the IEEE/CVF international conference on computer vision. 1314-1324.
29. [29] Tan, M., Le, Q. (2019). Efficientnet: Rethinking model scaling for convolutional neural networks. In International conference on machine learning. 6105-6114. PMLR.
30. [30] Dosovitskiy, A., Beyer, L., Kolesnikov, A., Weissenborn, D., Zhai, X., Unterthiner, T., Dehghani, M., Minderer, M., Heigold, G., Gelly, S., Uszkoreit, J. (2020). An image is worth 16x16 words: Transformers for image recognition at scale. arXiv preprint arXiv:2010.11929.
31. [31] Liu, Z., Lin, Y., Cao, Y., Hu, H., Wei, Y., Zhang, Z., Lin, S., Guo, B. (2021). Swin transformer: Hierarchical vision transformer using shifted windows. In Proceedings of the IEEE/CVF international conference on computer vision. 10012-10022.
32. [32] Radosavovic, I., Kosaraju, R. P., Girshick, R., He, K., Dollár, P. (2020). Designing network design spaces. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. 10428-10436.