Segmentasi Semantik Biner Banjir pada Citra UAV: Evaluasi Multi-Seed dan Studi Ablasi dengan Attention U-Net

Singgih Arya Mahendra; Ricardus Anggi Pramunendar

doi:10.59261/bustechno.v7i4.779

Authors

Singgih Arya Mahendra Universitas Dian Nuswantoro
Ricardus Anggi Pramunendar Universitas Dian Nuswantoro

DOI:

https://doi.org/10.59261/bustechno.v7i4.779

Keywords:

Segmentasi Semantik Biner, Deteksi Banjir, Attention U-Net, Focal Tversky Loss

Abstract

Latar belakang: Deteksi area banjir dari citra Unmanned Aerial Vehicle (UAV) menghadapi tantangan class imbalance, batas genangan yang tidak beraturan, dan biaya kesalahan yang asimetris antara false negative dan false positive. Pada situasi bencana, kegagalan mendeteksi area tergenang (false negative) berisiko jauh lebih fatal dibandingkan alarm palsu (false positive). Namun, model yang secara sistematis memprioritaskan recall sekaligus terbukti stabil secara berulang masih jarang dieksplorasi.

Tujuan: Penelitian ini mengusulkan pendekatan segmentasi semantik biner yang memadukan arsitektur Attention U-Net, backbone ResNet50 pra-latih, dan perhitungan Focal Tversky Loss (α = 0,3; β = 0,7; γ = 1,33) untuk menekan angka kegagalan deteksi.

Metode: Metode ini divalidasi pada Flood Semantic Segmentation Dataset (FSSD) melalui evaluasi berulang tiga kali (multi-seed) dan uji statistik studi ablasi 2x2.

Hasil: Hasilnya membuktikan kinerja yang tinggi dan stabil dengan F1-score 95,91%, IoU 93,67%, serta recall 97,09% yang konsisten melampaui presisi sebesar 95,36%, dengan simpangan baku F1-score maupun IoU yang sangat kecil yaitu 0,17%. Studi ablasi menegaskan bahwa Focal Tversky Loss (FTL) merupakan penggerak utama tingginya recall.

Kesimpulan: Metode usulan terbukti stabil dan dapat direproduksi pada berbagai proses pelatihan. Penelitian ini berkontribusi pada pengembangan kerangka segmentasi berorientasi recall yang divalidasi secara ketat—mengintegrasikan studi ablasi 2×2, evaluasi multi-seed, dan pengujian statistik—sehingga cocok untuk sistem peringatan bencana yang tidak mentoleransi kegagalan deteksi (false negative).

References

Abd Zaid, M. M., Mohammed, A. A., & Sumari, P. (2025). Remote sensing image classification using convolutional neural network (CNN) and transfer learning techniques. Journal of Computer Science, 21(3), 635–645. https://doi.org/10.3844/jcssp.2025.635.645

Abraham, N., & Khan, N. M. (2019). A Novel Focal Tversky Loss Function With Improved Attention U-Net for Lesion Segmentation. 2019 IEEE 16th International Symposium on Biomedical Imaging (ISBI 2019), 683–687. https://doi.org/10.1109/ISBI.2019.8759329

Büttner, M., Schneider, L., Krasowski, A., Pitchika, V., Krois, J., Meyer-Lueckel, H., & Schwendicke, F. (2024). Conquering class imbalances in deep learning-based segmentation of dental radiographs with different loss functions. Journal of Dentistry, 148, 105063. https://doi.org/10.1016/j.jdent.2024.105063

CRED, C. for R. on the E. of D. (2023). 2022 Disasters in numbers: Climate in action. Centre for Research on the Epidemiology of Disasters (CRED).

Erdelj, M., Natalizio, E., Chowdhury, K. R., & Akyildiz, I. F. (2017). Help from the sky: Leveraging UAVs for disaster management. IEEE Pervasive Computing, 16(1), 24–32. https://doi.org/10.1109/MPRV.2017.11

He, K., Zhang, X., Ren, S., & Sun, J. (2016). Deep Residual Learning for Image Recognition. 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 770–778. https://doi.org/10.1109/CVPR.2016.90

Hernández, D., Cecilia, J. M., Cano, J.-C., & Calafate, C. T. (2022). Flood detection using real-time image segmentation from unmanned aerial vehicles on edge-computing platform. Remote Sensing, 14(1), 223. https://doi.org/10.3390/rs14010223

Jiang, C., Wang, Y., Yuan, Q., Qu, P., & Li, H. (2025). A 3D medical image segmentation network based on gated attention blocks and dual-scale cross-attention mechanism. Scientific Reports, 15(1). https://doi.org/10.1038/s41598-025-90339-y

Kattenborn, T., Schiefer, F., Frey, J., Feilhauer, H., Mahecha, M. D., & Dormann, C. F. (2022). Spatially autocorrelated training and validation samples inflate performance assessment of convolutional neural networks. ISPRS Open Journal of Photogrammetry and Remote Sensing, 5, 100018. https://doi.org/10.1016/j.ophoto.2022.100018

Khan, B. A., & Jung, J.-W. (2024). Semantic segmentation of aerial imagery using U-Net with self-attention and separable convolutions. Applied Sciences, 14(9), 3712. https://doi.org/10.3390/app14093712

Lee, H., & Kang, M. G. (2023). Infrared Image Deconvolution Considering Fixed Pattern Noise. Sensors, 23(6). https://doi.org/10.3390/s23063033

Li, J., Hong, D., Gao, L., Yao, J., Zheng, K., Zhang, B., & Chanussot, J. (2022). Deep learning in multimodal remote sensing data fusion: A comprehensive review. International Journal of Applied Earth Observation and Geoinformation, 112. https://doi.org/10.1016/j.jag.2022.102926

Maxwell, A. E., Bester, M. S., & Ramezan, C. A. (2022). Enhancing reproducibility and replicability in remote sensing deep learning research and practice. Remote Sensing, 14(22), 5760. https://doi.org/10.3390/rs14225760

Mienye, I. D., Swart, T. G., Obaido, G., Jordan, M., & Ilono, P. (2025). Deep Convolutional Neural Networks in Medical Image Analysis: A Review. In Information (Switzerland) (Vol. 16, Number 3). https://doi.org/10.3390/info16030195

Milletari, F., Navab, N., & Ahmadi, S.-A. (2016). V-Net: Fully Convolutional Neural Networks for Volumetric Medical Image Segmentation. 2016 Fourth International Conference on 3D Vision (3DV), 565–571. https://doi.org/10.1109/3DV.2016.79

Odena, A., Dumoulin, V., & Olah, C. (2016). Deconvolution and checkerboard artifacts. Distill, 1(10). https://doi.org/10.23915/distill.00003

Rema, P., Moorthy, S., Bright, B., & BR, M. (2026). An Attention-Guided Residual 3D U-Net with Focal Tversky-Dice Loss for Multi-Modal Pancreatic Tumor Segmentation Using Synthetic Volumetric Imaging. Cancer Treatment and Research Communications, 101146.

Ronneberger, O., Fischer, P., & Brox, T. (2015). U-Net: Convolutional networks for biomedical image segmentation. In Medical Image Computing and Computer-Assisted Intervention – MICCAI 2015 (pp. 234–241). Springer. https://doi.org/10.1007/978-3-319-24574-4_28

Salehi, S. S. M., Erdogmus, D., & Gholipour, A. (2017). Tversky loss function for image segmentation using 3D fully convolutional deep networks. In Machine Learning in Medical Imaging (MLMI 2017) (Vol. 10541, pp. 379–387). Springer. https://doi.org/10.1007/978-3-319-67389-9_44

Schlemper, J., Oktay, O., Schaap, M., Heinrich, M., Kainz, B., Glocker, B., & Rueckert, D. (2019). Attention gated networks: Learning to leverage salient regions in medical images. Medical Image Analysis, 53, 197–207. https://doi.org/10.1016/j.media.2019.01.012

Shafiq, M., & Gu, Z. (2022). Deep Residual Learning for Image Recognition: A Survey. In Applied Sciences (Switzerland) (Vol. 12, Number 18). https://doi.org/10.3390/app12188972

Simantiris, G., & Panagiotakis, C. (2024). Unsupervised color-based flood segmentation in UAV imagery. Remote Sensing, 16(12), 2126. https://doi.org/10.3390/rs16122126

Sundaresan, A. A., & Solomon, A. A. (2025). Post-disaster flooded region segmentation using DeepLabv3+ and unmanned aerial system imagery. Natural Hazards Research, 5(2), 363–371. https://doi.org/10.1016/j.nhres.2024.12.003

Thapa, A., Horanont, T., Neupane, B., & Aryal, J. (2023). Deep learning for remote sensing image scene classification: A review and meta-analysis. Remote Sensing, 15(19), 4804. https://doi.org/10.3390/rs15194804

Wibowo, A., Amri, I., Surahmat, A., & Rusdah, R. (2025). Leveraging artificial intelligence in disaster management: A comprehensive bibliometric review. Jamba: Journal of Disaster Risk Studies, 17(1). https://doi.org/10.4102/jamba.v17i1.1776

Yeung, M., Sala, E., Schönlieb, C.-B., & Rundo, L. (2022). Unified focal loss: Generalising Dice and cross entropy-based losses to handle class imbalanced medical image segmentation. Computerized Medical Imaging and Graphics, 95, 102026. https://doi.org/10.1016/j.compmedimag.2021.102026

Zhang, J., Zhang, Y., Jin, Y., Xu, J., & Xu, X. (2023). MDU-Net: multi-scale densely connected U-Net for biomedical image segmentation. Health Information Science and Systems, 11(1). https://doi.org/10.1007/s13755-022-00204-9

Zhuang, F., Qi, Z., Duan, K., Xi, D., Zhu, Y., Zhu, H., Xiong, H., & He, Q. (2021). A comprehensive survey on transfer learning. Proceedings of the IEEE, 109(1), 43–76. https://doi.org/10.1109/JPROC.2020.3004555

Zhu, R., Oda, M., Hayashi, Y., Kitasaka, T., Misawa, K., Fujiwara, M., & Mori, K. (2025). Skeleton-guided 3D convolutional neural network for tubular structure segmentation. International Journal of Computer Assisted Radiology and Surgery, 20(1). https://doi.org/10.1007/s11548-024-03215-x

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. https://doi.org/10.1109/MGRS.2017.2762307

Wallemacq, P., & House, R. (2018). Economic losses, poverty & disasters: 1998-2017. Centre for Research on the Epidemiology of Disasters (CRED); United Nations Office for Disaster Risk Reduction (UNDRR). https://www.undrr.org/publication/economic-losses-poverty-disasters-1998-2017