Cybernetics And Systems Analysis logo
Editorial Board Announcements Abstracts Authors Archive
KIBERNETYKA TA SYSTEMNYI ANALIZ
International Theoretical Science Journal
-->

DOI 10.34229/KCA2522-9664.24.5.3
UDC 004.8, 004.93
A. Okhrimenko1, N. Kussul2


1 National Technical University of Ukraine “Igor Sikorsky Kyiv Polytechnic Institute,” Kyiv, Ukraine

antoh-ipt21@lll.kpi.ua

2 National Technical University of Ukraine “Igor Sikorsky Kyiv Polytechnic Institute,” Kyiv, Ukraine

nataliia.kussul@gmail.com

USING WEIGHT RELIABILITY MASKS ON IMBALANCED DATASETS
FOR SATELLITE IMAGE SEGMENTATION

Abstract. This study addresses the issue of imbalanced datasets in satellite image segmentation tasks, where machine learning models often neglect minority classes in favor of majority ones. We propose using spatial weight masks for the loss function computation to take into account the reliability score of individual pixels. This approach enhances segmentation quality, significantly improving metrics for minority classes. Additionally, a dataset augmentation method using generative adversarial networks (GANs) is explored, showing slight improvements in recognizing less represented crop types in the dataset, and it is compared with the proposed method. The simultaneous usage of weighted masks and generative networks is investigated.

Keywords: dataset quality assessment, imbalanced datasets, classification, segmentation, generative adversarial networks, training data generation.


full text

REFERENCES

  • 1. Banko M., Brill E. Scaling to very very large corpora for natural language disambiguation. Proc. 39th Annual Meeting on Association for Computational Linguistics (ACL’ 01) ( 6–11 July 2001, Toulouse, France). Toulouse, 2001. P. 26–33. doi.org/10.3115/1073012.1073017.

  • 2. Wang L., Han M., Li X., Zhang N., Cheng H. Review of classification methods on unbalanced data sets. IEEE Access. 2021. Vol. 9. P. 64606–64628. doi.org/10.1109/ACCESS.2021.3074243.

  • 3. Kumar A., Goel S., Sinha N., Bhardwaj A. A review on unbalanced data classification. Proc. International Joint Conference on Advances in Computational Intelligence (IJCACI 2021) (23–24 October 2021, online event). P. 197–208. doi.org/10.1007/978-981-19-03328_14.

  • 4. Li X., Zhang L. Unbalanced data processing using deep sparse learning technique. Futur. Gener. Comput. Syst. 2021. Vol. 125, P. 480–484. doi.org/10.1016/j.future.2021.05.034.

  • 5. Viloria A., Lezama O.B.P., Mercado-Caruzo N. Unbalanced data processing using oversampling: Machine learning. Procedia Computer Science. 2020. Vol. 175. P.108–113. doi.org/10.1016/j.procs.2020.07.018.

  • 6. Hido S., Kashima H., Takahashi Y. Roughly balanced bagging for imbalanced data. Stat. Anal. Data Min. 2009. Vol. 2, Iss. 5–6. P. 412–426. doi.org/10.1002/sam.10061.

  • 7. Lango M., Stefanowski J. Multi-class and feature selection extensions of Roughly Balanced Bagging for imbalanced data. J. Intell. Inf. Syst. 2018. Vol. 50, N 1. P. 97–127. doi.org/10.1007/s10844-017-0446-7.

  • 8. Lssig N., Oppold S., Herschel M. Metrics and algorithms for locally fair and accurate classifications using ensembles. Datenbank-Spektrum. 2022. Vol. 22, N 1. P. 23–43. doi.org/10.1007/s13222-021-00401-y .

  • 9. Tang Y., Gao J. Improved classification for problem involving overlapping patterns. IEICE Trans. Inf. Syst. 2007. Vol. E90-D, N 11. P. 1787–1795. doi.org/10.1093/ietisy/e90-d.11.1787.

  • 10. Cui Y., Jia M., Lin T.Y., Song Y., Belongie S. Class-balanced loss based on effective number of samples. Proc. 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (15-20 June 2019, Long Beach, CA, USA). Long Beach, 2019. P. 9260–9269. doi.org/10.1109/CVPR.2019.00949.

  • 11. Phan T.H., Yamamoto K. Resolving class imbalance in object detection with weighted cross entropy losses. arXiv:2006.01413v1 [cs.CV] 2 Jun 2020. doi.org/10.48550/arXiv.2006.01413.

  • 12. Qiao X., Liu Y. Adaptive weighted learning for unbalanced multicategory classification. Biometrics. 2009. Vol. 65, Iss. 1. P. 159–168. doi.org/10.1111/j.1541-0420.2008.01017.x .

  • 13. Shorten C., Khoshgoftaar T.M. A survey on image data augmentation for deep learning. J. Big Data. 2019. Vol. 6, N 1. Article number 60. doi.org/10.1186/s40537-019-0197-0.

  • 14. Abady L., Barni M., Garzelli A., Tondi B. GAN generation of synthetic multispectral satellite images. Proc. Image and Signal Processing for Remote Sensing XXVI (21–25 September 2020, online event, UK). Vol. 11533. 115330L. doi.org/10.1117/12.2575765.

  • 15. Shah M., Gupta M., Thakkar P. SatGAN: Satellite image generation using conditional adversarial networks. Proc. 2021 International Conference on Communication, Information and Computing Technology (ICCICT 2021) (25-27 June 2021, Mumbai, India). Mumbai, 2021. P. 1–6. doi.org/10.1109/ICCICT50803.2021.9510104.

  • 16. Hu W., Yin Y., Tan Y.K., Tran A., Kruppa H., Zimmermann R. GAN-assisted road segmentation from satellite imagery. ACM Trans. Multimed. Comput. Commun. Appl. 2023. Just accepted. doi.org/10.1145/3635153.

  • 17. Shumilo L., Okhrimenko A., Kussul N., Drozd S., Shkalikov O. Generative adversarial network augmentation for solving the training data imbalance problem in crop classification. Remote Sens. Lett. 2023. Vol. 14, N 11. P. 1131–1140. doi.org/10.1080/2150704X.2023.2275551.

  • 18. Lavreniuk M., Kussul N., Novikov A. Deep learning crop classification approach based on sparse coding of time series of satellite data. Proc. IGARSS 2018 - 2018 IEEE International Geoscience and Remote Sensing Symposium (22–27 July 2018, Valencia, Spain). Valencia, 2018. P. 4812–4815. doi.org/10.1109/IGARSS.2018.8518263.

  • 19. Okhrimenko A.O., Kussul N.M. A method for identifying difficult-to-recognize patterns in datasets for classification tasks in machine learning. Problems of control and informatics. 2023. Vol. 68, N. 4. P. 84–95. doi.org/10.34229/1028-0979-2023-4-7.

  • 20. Ronneberger O., Fischer P., Brox T. U-Net: convolutional networks for biomedical image segmentation. Proc. Medical Image Computing and Computer-Assisted Intervention (MICCAI 2015) (5–9 October 2015, Munich, Germany). Munich, 2015. LNCS. 2015. Vol. 9351, P. 234–241. doi.org/10.1007/978-3-319-24574-4_28.




© 2024 Kibernetika.org. All rights reserved.