Derin Öğrenme Yaklaşımlarıyla Çevresel İzlemeye Yönelik Çok-Sınıflı Sınıflandırma

Öz

Sınıflandırma haritaları, çevresel izleme görevlerinin ana çıktı türlerinden biridir. Bu çalışmada, görüntü sınıflandırması için uzaktan algılama verileri kullanılarak derin öğrenme algoritmaları uygulanmıştır. Uygulamada UC Merced ve WHU-RS19 olmak üzere iki veri seti üzerinde farklı CNN modelleri kullanılmıştır. Test aşamasında derin öğrenme modellerinin tahminleri ile çok-sınıflı sınıflandırma yapılmış ve sınıflandırmaya ait değerlendirme ölçütleri hesaplanmıştır. Kullanılan CNN modellerinin veri setlerindeki performansları genel doğruluk ölçütünde değerlendirilmiştir. DenseNet201 modelinin, UC Merced ve WHU-RS19 veri setlerinin her ikisinde de testlerde daha yüksek performanslı sonuçlara sahip olduğu gözlemlenmiştir. Elde edilen sonuçlar, literatürdeki diğer çalışmaların sonuçlarıyla karşılaştırılmıştır. UC Merced veri setindeki uygulamada %98.81 genel doğruluk ile bu çalışmada kullanılan DenseNet201 modelinin diğer çalışmalardan daha yüksek performansa sahip olduğu gözlenmiştir. Ayrıca, her iki veri setinde benzer olan arazi kullanım sınıfları belirlenmiş ve en iyi performans gösteren algoritmadaki sonuçları yorumlanmıştır, Benzer sınıfların yapılan testlerde sınıflandırılması kesinlik, duyarlılık ve F1 skoru ölçütleri kullanılarak değerlendirilmiştir.

Anahtar Kelimeler

• Derin öğrenme
• Çevresel izleme
• Evrişimsel Sinir Ağları
• Çok-sınıflı sınıflandırma

Multi-Class Classification for Environmental Monitoring with Deep Learning Approaches

Öz

Classification maps are one of the main output types of environmental monitoring tasks. In this study, deep learning algorithms were applied using remote sensing data for image classification. In the application, different CNN models were used on two data sets, UC Merced and WHU-RS19. In the test phase, multi-class classification was made with the predictions of deep learning models and the evaluation criteria of the classification were calculated. The performances of the CNN models used in the data sets were evaluated in the overall accuracy metric. It has been observed that the DenseNet201 model has higher performance results in tests on both the UC Merced and WHU-RS19 datasets. The results obtained were compared with the results of other studies in the literature. It has been observed that the DenseNet201 model used in this study has higher performance than other studies with an overall accuracy of 98.81% in the application in the UC Merced dataset. In addition, land use classes that are similar in both data sets were determined and the results of the best performing algorithm were interpreted. Classification of similar classes in the tests was evaluated using the evaluation metrics of precision, recall and F1 score.

Anahtar Kelimeler

• Deep learning
• Environmental Monitoring
• Convolutional Neural Networks; Multi-class classification

Kaynakça

1. Aksoy, A. K., Ravanbakhsh, M., Kreuziger, T., & Demir, B. (2020). CCML: A Novel Collaborative Learning Model for Classification of Remote Sensing Images with Noisy Multi-Labels. arXiv preprint arXiv:2012.10715.
2. Anwer, R. M., Khan, F. S., van de Weijer, J., Molinier, M., & Laaksonen, J. (2018). Binary patterns encoded convolutional neural networks for texture recognition and remote sensing scene classification. ISPRS journal of photogrammetry and remote sensing, 138, 74-85.
3. Atik, M. E., Donmez, S. O., Duran, Z., & İpbüker, C. (2018). Comparison Of Automatic Feature Extraction Methods For Building Roof Planes By Using Airborne Lidar Data And High Resolution Satellite Image. Proceeding Book of 7th International Conference on Cartography and GIS, 18-23 June 2018, Sozopol, Bulgaria.
4. Atik, S. O., & Ipbuker, C. (2021). Integrating Convolutional Neural Network and Multiresolution Segmentation for Land Cover and Land Use Mapping Using Satellite Imagery. Applied Sciences, 11(12), 5551.
5. Atik, S.O., Ipbuker, C. (2020) Instance segmentation of crowd detection in the camera images, In Proceedings of the 41th Asian Conference on Remote Sensing (ACRS), Deqing, China, 9-11 November 2020.
6. Bi, Q., Qin, K., Zhang, H., Xie, J., Li, Z., & Xu, K. (2019). APDC-Net: Attention pooling-based convolutional network for aerial scene classification. IEEE Geoscience and Remote Sensing Letters, 17(9), 1603-1607.
7. Brown, M., & Süsstrunk, S. (2011, June). Multi-spectral SIFT for scene category recognition. In CVPR 2011 (pp. 177-184). IEEE.
8. 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.

9. Diakogiannis, F. I., Waldner, F., Caccetta, P., & Wu, C. (2020). ResUNet-a: A deep learning framework for semantic segmentation of remotely sensed data. ISPRS Journal of Photogrammetry and Remote Sensing, 162, 94-114.
10. Donmez, S.O.; Ipbuker, C. Investigation on Agent Based Models for Image Classification of Land Use and Land Cover Maps. In Proceedings of the 39th Asian Conference on Remote Sensing (ACRS): Remote Sensing Enabling Prosperity, Kuala Lumpur, Malaysia, 15–19 October 2018; pp. 2005–2008.
11. Dönmez, Ş. Ö., & Tunc, A. (2016). Transformation methods for using combination of remotely sensed data and cadastral maps. The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, 4, 587-589.
12. 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 (pp. 770-778).
13. Hu, J., Shu, Q., Pan, J., Tu, J., Zhu, Y., & Wang, M. (2021). MINet: Multilevel Inheritance Network-Based Aerial Scene Classification. IEEE Geoscience and Remote Sensing Letters, 19, 1-5.
14. 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 (pp. 4700-4708).
15. Krizhevsky, A., Sutskever, I., & Hinton, G. E. (2012). Imagenet classification with deep convolutional neural networks. Advances in neural information processing systems, 25, 1097-1105.
16. Marmanis, D., Datcu, M., Esch, T., & Stilla, U. (2015). Deep learning earth observation classification using ImageNet pretrained networks. IEEE Geoscience and Remote Sensing Letters, 13(1), 105-109.
17. Muhammad, U., Wang, W., Chattha, S. P., & Ali, S. (2018, August). Pre-trained VGGNet architecture for remote-sensing image scene classification. In 2018 24th International Conference on Pattern Recognition (ICPR) (pp. 1622-1627). IEEE.
18. Napiorkowska, M., Petit, D., & Marti, P. (2018, July). Three applications of deep learning algorithms for object detection in satellite imagery. In IGARSS 2018-2018 IEEE International Geoscience and Remote Sensing Symposium (pp. 4839-4842). IEEE.
19. Negrel, R., Picard, D., & Gosselin, P. H. (2014, June). Evaluation of second-order visual features for land-use classification. In 2014 12th International Workshop on Content-Based Multimedia Indexing (CBMI) (pp. 1-5). IEEE.
20. Özyurt, F., Ava, E., & Sert, E. (2020). UC-merced image classification with cnn feature reduction using wavelet entropy optimized with genetic algorithm.
21. Penatti, O. A., Nogueira, K., & Dos Santos, J. A. (2015). Do deep features generalize from everyday objects to remote sensing and aerial scenes domains?. In Proceedings of the IEEE conference on computer vision and pattern recognition workshops (pp. 44-51).
22. Petrovska, B., Zdravevski, E., Lameski, P., Corizzo, R., Štajduhar, I., & Lerga, J. (2020). Deep learning for feature extraction in remote sensing: A case-study of aerial scene classification. Sensors, 20(14), 3906.
23. Qi, X., Zhu, P., Wang, Y., Zhang, L., Peng, J., Wu, M., ... & Mathiopoulos, P. T. (2020). MLRSNet: A multi-label high spatial resolution remote sensing dataset for semantic scene understanding. ISPRS Journal of Photogrammetry and Remote Sensing, 169, 337-350.
24. Sheng, G., Yang, W., Xu, T., & Sun, H. (2012). High-resolution satellite scene classification using a sparse coding based multiple feature combination. International journal of remote sensing, 33(8), 2395-2412.
25. Simonyan, K., & Zisserman, A. (2014). Very deep convolutional networks for large-scale image recognition. arXiv preprint arXiv:1409.1556.
26. Stivaktakis, R., Tsagkatakis, G., & Tsakalides, P. (2019). Deep learning for multilabel land cover scene categorization using data augmentation. IEEE Geoscience and Remote Sensing Letters, 16(7), 1031-1035.
27. Sumbul, G., Charfuelan, M., Demir, B., & Markl, V. (2019, July). Bigearthnet: A large-scale benchmark archive for remote sensing image understanding. In IGARSS 2019-2019 IEEE International Geoscience and Remote Sensing Symposium (pp. 5901-5904). IEEE.
28. Weng, Q., Mao, Z., Lin, J., & Guo, W. (2017). Land-use classification via extreme learning classifier based on deep convolutional features. IEEE Geoscience and Remote Sensing Letters, 14(5), 704-708.
29. Xia, G. S., Hu, J., Hu, F., Shi, B., Bai, X., Zhong, Y., ... & Lu, X. (2017). AID: A benchmark data set for performance evaluation of aerial scene classification. IEEE Transactions on Geoscience and Remote Sensing, 55(7), 3965-3981.
30. Yang, Y., & Newsam, S. (2010, November). Bag-of-visual-words and spatial extensions for land-use classification. In Proceedings of the 18th SIGSPATIAL international conference on advances in geographic information systems (pp. 270-279).
31. Zhou, W., Newsam, S., Li, C., & Shao, Z. (2018). PatternNet: A benchmark dataset for performance evaluation of remote sensing image retrieval. ISPRS journal of photogrammetry and remote sensing, 145, 197-209.