Segmentation of River Networks from Multispectral Remote Sensing Data Using Deep Neural Networks

• Eduardo H. P. Souza IFES
• Francisco de A. Boldt IFES
• Thiago M. Paixão IFES
• Jefferson O. Andrade IFES
• Karin S. Komati IFES

DOI: https://doi.org/10.5753/sbsi.2026.248616

Resumo

Research Context: Accurate hydrological mapping is essential for water resource management, flow modeling, and risk assessment of flooding and erosion. Scientific Problem: Conventional methods and RGB satellite imagery fail to map watercourses obscured by dense vegetation, resulting in fragmented, unreliable cartography that compromises territorial planning. Proposed Solution: We propose a comparative analysis of river network semantic segmentation, evaluating U-Net and DeepLabv3+ architectures with ImageNet pre-trained backbones (EfficientNet-B5, ResNet-152). The analysis utilizes multispectral data (RGB, NIR) and spectral indices (NDWI, NDVI, GNDVI) to enhance feature discrimination. Related IS Theory: This work aligns with TaskTechnology Fit by evaluating the suitability of deep learning for automated river network extraction. Research Method: We created a dataset from satellite imagery and official vector data of two Brazilian basins (Doce, Itapemirim). Models were trained in intra-dataset, combined, and cross-dataset scenarios using mean Intersection over Union (mIoU). Summary of Results: For intra-dataset evaluation, U-Net using EfficientNet-B5 yielded 84.86±0.39% mIoU, whereas DeepLabv3+ with ResNet-152 showed 86.73±0.21%. Combined datasets provided stable metrics (82.25±0.51% for U-Net). Cross-dataset tests reached 63.96% mIoU, indicating a performance decrease. Contributions/Impact: This study demonstrates that U-Net and DeepLabv3+ architectures using deep backbones accurately segment river networks in complex environments. The methodology provides a viable approach to automate drainage network extraction, supporting environmental management and hydrological modeling.

Referências

Bardis, M., Houshyar, R., Chantaduly, C., Ushinsky, A., Glavis-Bloom, J., Shaver, M., Chow, D., Uchio, E., and Chang, P. (2020). Deep learning with limited data: organ segmentation performance by U-Net. Electronics, 9(8):1199.

Biradar, R. L., Thatipalli, S., Mucharla, A., Adepu, S., and Mandava, P. (2024). Detection of water bodies using satellite imagery based on deep learning. International Journal for Research in Applied Science & Engineering Technology (IJRASET), 12(5).

Blais, M.-A. and Akhloufi, M. A. (2025). Benchmarking coastal boundary datasets in deep learning applications. Earth Science Informatics, 18(4):520.

Cao, H., Tian, Y., Liu, Y., and Wang, R. (2024). Water body extraction from high spatial resolution remote sensing images based on enhanced U-Net and multi-scale information fusion. Scientific Reports, 14(1):16132.

Chen, L.-C., Zhu, Y., Papandreou, G., Schroff, F., and Adam, H. (2018). Encoder-decoder with atrous separable convolution for semantic image segmentation. In Proceedings of the European conference on computer vision (ECCV), pages 801–818.

Everingham, M., Eslami, S. M. A., Gool, L. V., Williams, C. K. I., Winn, J., and Zisserman, A. (2010). The pascal visual object classes (VOC) challenge. International Journal of Computer Vision, 88(2):303–338.

Fawzy, M. and Barsi, A. (2025). A U-Net model for urban land cover classification using vhr satellite images. Periodica Polytechnica Civil Engineering, 69(1):98–108.

Fei-Fei, L., Deng, J., and Li, K. (2009). Imagenet: Constructing a large-scale image database. Journal of vision, 9(8):1037–1037.

Feng, S. J., Feng, Y., Zhang, X. L., and Chen, Y. H. (2023). Deep learning with visual explanations for leakage defect segmentation of metro shield tunnel. Tunnelling and Underground Space Technology, 136:105107.

Ganin, Y., Ustinova, E., Ajakan, H., Germain, P., Larochelle, H., Laviolette, F., Marchand, M., and Lempitsky, V. (2016). Domain-adversarial training of neural networks. The journal of machine learning research, 17(1):2096–2030.

Gitelson, A. A., Kaufman, Y. J., and Merzlyak, M. N. (1996). Use of a green channel in remote sensing of global vegetation from EOS-MODIS. Remote sensing of Environment, 58(3):289–298.

He, K., Zhang, X., Ren, S., and Sun, J. (2016). Deep residual learning for image recognition. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 770–778.

McFeeters, S. K. (1996). The use of the normalized difference water index (NDWI) in the delineation of open water features. International journal of remote sensing, 17(7):1425–1432.

Milletari, F., Navab, N., and Ahmadi, S.-A. (2016). V-Net: fully convolutional neural networks for volumetric medical image segmentation. arXiv preprint arXiv:1606.04797, pages 565–571.

Osias, A. C. F., Schaefer, M. A. R., Veloso, G. V., de Oliveira, H. N., and Reis, J. C. S. (2024). Interpretable approaches for land use and land cover classification. In Proceedings of Simpósio Brasileiro de Sistemas de Informação (SBSI’24), SBSI’24, New York, NY, USA. Association for Computing Machinery.

Patil, P. P., Jagtap, M. P., Khatri, N., Madan, H., Vadduri, A. A., and Patodia, T. (2024). Exploration and advancement of NDDI leveraging NDVI and NDWI in Indian semi-arid regions: A remote sensing-based study. Case Studies in Chemical and Environmental Engineering, 9:100573.

QGIS Development Team (2025). QGIS Geographic Information System. Open Source Geospatial Foundation. Version 3.36.

Ronneberger, O., Fischer, P., and Brox, T. (2015). U-Net: convolutional networks for biomedical image segmentation. In Medical Image Computing and Computer-Assisted Intervention – MICCAI 2015, pages 234–241. Springer.

Rouse, J. W., Haas, R. H., Schell, J. A., and Deering, D. W. (1974). Monitoring vegetation systems in the great plains with erts. NASA special publication, 351:309.

Rusnák, M., Goga, T., Michaleje, L., Šulc Michalková, M., Máčka, Z., Bertalan, L., and Kidová, A. (2022). Remote sensing of riparian ecosystems. Remote Sensing, 14(11):2645.

Shen, J., Guo, Z., Zhang, Z., Plathong, S., Jantharakhantee, C., Ma, J., Ning, H., and Qi, Y. (2025). Remote sensing shoreline extraction method based on an optimized deeplabv3+ model: A case study of koh lan island, thailand. Journal of Marine Science and Engineering, 13(4):665.

Souza, E. H. P., Oliveira, V. M. d., Andrade, J. O., and Komati, K. S. (2025). Previsão de vazão de rios usando rede perceptron multi-camada otimizada por neural architecture search. Tecnia: Revista de Educação, Ciência e Tecnologia do IFG, 10(Edição Especial 1).

Sun, D., Gao, G., Huang, L., Liu, Y., and Liu, D. (2024). Extraction of water bodies from high-resolution remote sensing imagery based on a deep semantic segmentation network. Scientific Reports, 14(1):14604.

Tan, M. and Le, Q. V. (2019). EfficientNet: rethinking model scaling for convolutional neural networks. In Chaudhuri, K. and Salakhutdinov, R., editors, Proceedings of the 36th International Conference on Machine Learning (ICML), volume 97 of Proceedings of Machine Learning Research, pages 6105–6114. PMLR.

Tiwari, T. and Saraswat, M. (2024). Analysis of UNet-Based semantic segmentation models. In International Conference on Computing and Machine Learning, pages 421–431. Springer.

Viana, A. B., da Silva, R. A., and de Oliveira, V. d. P. S. (2024). Uso racional de água de reuso ou potável na indústria. Boletim do Observatório Ambiental Alberto Ribeiro Lamego, 18(2):17–35.

Wang, Y., He, J., Wang, C., and Zhang, W. (2026). Wetland information extraction method based on improved deeplabv3+ in liaohe river estuary. Wetlands, 46(1):11.

Weng, L., Pang, K., Xia, M., Lin, H., Qian, M., and Zhu, C. (2023). Sgformer: A local and global features coupling network for semantic segmentation of land cover. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 16:6812–6824.

Yap, H. Y., Choo, Y.-H., Mohd Yusoh, Z. I., and Khoh, W. H. (2023). An evaluation of transfer learning models in EEG-based authentication. Brain informatics, 10(1):19.

Yosinski, J., Clune, J., and O. L. Sinsap, Yoshua Bengio, H. L. (2014). How transferable are features in deep neural networks? In Advances in neural information processing systems, volume 27.

Yuan, K., Zhuang, X., Schaefer, G., Feng, J., Guan, L., and Fang, H. (2021). Deep-learning-based multispectral satellite image segmentation for water body detection. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 14:7422–7434.

Zhang, Y., Yang, R., Dai, Q., Zhao, Y., Xu, W., Wang, J., and Wang, L. (2023). Boosting semantic segmentation of remote sensing images by introducing edge extraction network and spectral indices. Remote Sensing, 15(21):5148.