Reducing Manual Workload in SAR-Based Oil Spill Detection Through Uncertainty-Aware Deep Learning

Dina Svendsen Solskinnsbakk, Sigurd Almli Hanssen, Harald Lykke Joakimsen, Vilde B. Gjærum, Elisabeth Wetzer, Kristoffer Knutsen Wickstrøm
Proceedings of the 7th Northern Lights Deep Learning Conference (NLDL), PMLR 307:364-374, 2026.

Abstract

Constant monitoring of the oceans is required to detect oil spills and reduce environmental damage associated with spills. Synthetic Aperture Radar (SAR) imaging is a critical tool for oil spill detection, but is complex and requires time-consuming manual labor for analysis. Deep learning has shown encouraging performance in automatic classification of oil spills on these images, but the performance is still not sufficient for a deep learning classifier to act autonomously, making manual assessment essential. However, if only a reduced subset of uncertain samples had to be analyzed by human experts while the remaining samples could be automatically classified, it could greatly reduce the manual workload. In this study, we investigate if uncertainty estimates can identify which samples should be prioritized for manual inspection. Specifically, we propose a pipeline of defining a user-specified error tolerance and identifying an uncertainty threshold that filters out samples for automatic/manual thresholding. We evaluate the proposed pipeline on challenging real-world data. The results show that our proposed uncertainty-based ranking technique can reduce the manual workload by 41%, paving the way for new and more efficient ways to detect marine oil spills.

Cite this Paper

BibTeX
@InProceedings{pmlr-v307-solskinnsbakk26a, title = {Reducing Manual Workload in {SAR}-Based Oil Spill Detection Through Uncertainty-Aware Deep Learning}, author = {Solskinnsbakk, Dina Svendsen and Hanssen, Sigurd Almli and Joakimsen, Harald Lykke and Gj{\ae}rum, Vilde B. and Wetzer, Elisabeth and Wickstr{\o}m, Kristoffer Knutsen}, booktitle = {Proceedings of the 7th Northern Lights Deep Learning Conference (NLDL)}, pages = {364--374}, year = {2026}, editor = {Kim, Hyeongji and Ramírez Rivera, Adín and Ricaud, Benjamin}, volume = {307}, series = {Proceedings of Machine Learning Research}, month = {06--08 Jan}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v307/main/assets/solskinnsbakk26a/solskinnsbakk26a.pdf}, url = {https://proceedings.mlr.press/v307/solskinnsbakk26a.html}, abstract = {Constant monitoring of the oceans is required to detect oil spills and reduce environmental damage associated with spills. Synthetic Aperture Radar (SAR) imaging is a critical tool for oil spill detection, but is complex and requires time-consuming manual labor for analysis. Deep learning has shown encouraging performance in automatic classification of oil spills on these images, but the performance is still not sufficient for a deep learning classifier to act autonomously, making manual assessment essential. However, if only a reduced subset of uncertain samples had to be analyzed by human experts while the remaining samples could be automatically classified, it could greatly reduce the manual workload. In this study, we investigate if uncertainty estimates can identify which samples should be prioritized for manual inspection. Specifically, we propose a pipeline of defining a user-specified error tolerance and identifying an uncertainty threshold that filters out samples for automatic/manual thresholding. We evaluate the proposed pipeline on challenging real-world data. The results show that our proposed uncertainty-based ranking technique can reduce the manual workload by 41%, paving the way for new and more efficient ways to detect marine oil spills.} }
Endnote
%0 Conference Paper %T Reducing Manual Workload in SAR-Based Oil Spill Detection Through Uncertainty-Aware Deep Learning %A Dina Svendsen Solskinnsbakk %A Sigurd Almli Hanssen %A Harald Lykke Joakimsen %A Vilde B. Gjærum %A Elisabeth Wetzer %A Kristoffer Knutsen Wickstrøm %B Proceedings of the 7th Northern Lights Deep Learning Conference (NLDL) %C Proceedings of Machine Learning Research %D 2026 %E Hyeongji Kim %E Adín Ramírez Rivera %E Benjamin Ricaud %F pmlr-v307-solskinnsbakk26a %I PMLR %P 364--374 %U https://proceedings.mlr.press/v307/solskinnsbakk26a.html %V 307 %X Constant monitoring of the oceans is required to detect oil spills and reduce environmental damage associated with spills. Synthetic Aperture Radar (SAR) imaging is a critical tool for oil spill detection, but is complex and requires time-consuming manual labor for analysis. Deep learning has shown encouraging performance in automatic classification of oil spills on these images, but the performance is still not sufficient for a deep learning classifier to act autonomously, making manual assessment essential. However, if only a reduced subset of uncertain samples had to be analyzed by human experts while the remaining samples could be automatically classified, it could greatly reduce the manual workload. In this study, we investigate if uncertainty estimates can identify which samples should be prioritized for manual inspection. Specifically, we propose a pipeline of defining a user-specified error tolerance and identifying an uncertainty threshold that filters out samples for automatic/manual thresholding. We evaluate the proposed pipeline on challenging real-world data. The results show that our proposed uncertainty-based ranking technique can reduce the manual workload by 41%, paving the way for new and more efficient ways to detect marine oil spills.
APA
Solskinnsbakk, D.S., Hanssen, S.A., Joakimsen, H.L., Gjærum, V.B., Wetzer, E. & Wickstrøm, K.K.. (2026). Reducing Manual Workload in SAR-Based Oil Spill Detection Through Uncertainty-Aware Deep Learning. Proceedings of the 7th Northern Lights Deep Learning Conference (NLDL), in Proceedings of Machine Learning Research 307:364-374 Available from https://proceedings.mlr.press/v307/solskinnsbakk26a.html.

Related Material