Comparative Analysis of Activation Functions in U-Net for Binary Water Segmentation using Sentinel-2 Imagery
| dc.contributor.author | Kundenko, Pavlo | en |
| dc.contributor.author | Hnatushenko, Viktoriia V. | en |
| dc.contributor.author | Tsaryk, Vladyslav Yu. | en |
| dc.contributor.author | Dmytriieva, Iryna S. | en |
| dc.date.accessioned | 2025-07-14T08:09:17Z | |
| dc.date.issued | 2025 | |
| dc.description | P. Kundenko: ORCID 0009-0000-9388-2783; Vic. Hnatushenko: ORCID 0000-0001-5304-4144; V. Tsaryk: ORCID 0000-0001-6449-8037; I. Dmytriieva: ORCID 0009-0008-3298-7563 | en |
| dc.description.abstract | ENG: The study examines how different activation functions influence the performance of a U-Net model applied to binary water-body segmentation in Sentinel-2 imagery. Using an identical training setup for each experiment, six nonlinearities—ReLU, Leaky ReLU, ELU, PReLU, Swish and RReLU—are individually substituted into the network while all other parameters remain fixed. Comparative evaluation on a held-out validation set reveals that Leaky ReLU provides the most balanced trade-off between precision and recall, making it the preferred choice for accurate water-mask generation. PReLU offers a similar but slightly lower performance, whereas ELU excels at capturing additional water pixels at the cost of more false positives. The findings highlight the importance of activation-function selection in remote-sensing segmentation tasks and suggest further exploration of advanced nonlinearities and larger, more diverse datasets to enhance generalization. | en |
| dc.identifier.citation | Kundenko P., Hnatushenko Vic., Tsaryk V., Dmytriieva I. Comparative Analysis of Activation Functions in U-Net for Binary Water Segmentation using Sentinel-2 Imagery. CEUR Workshop Proceedings. Vol. 3983 : Proc. of the Intelligent Systems Workshop at 9th International Conference on Computational Linguistics and Intelligent Systems (CoLInS-2025), Kharkiv, Ukraine, May 15-16, 2025. Kharkiv, 2025. P. 142–152. DOI: https://doi.org/10.31110/COLINS/2025-2/011. | en |
| dc.identifier.doi | https://doi.org/10.31110/COLINS/2025-2/011 | en |
| dc.identifier.issn | 1613-0073 | |
| dc.identifier.uri | https://ceur-ws.org/Vol-3983/ | en |
| dc.identifier.uri | https://crust.ust.edu.ua/handle/123456789/20788 | en |
| dc.language.iso | en | |
| dc.publisher | CEUR-WS Team, Aachen, Germany | en |
| dc.subject | water-body segmentation | en |
| dc.subject | Sentinel-2 | en |
| dc.subject | U-Net | en |
| dc.subject | activation functions | en |
| dc.subject | remote sensing | en |
| dc.subject | deep learning | en |
| dc.subject | КІТС | uk_UA |
| dc.subject.classification | TECHNOLOGY::Information technology::Image analysis | en |
| dc.title | Comparative Analysis of Activation Functions in U-Net for Binary Water Segmentation using Sentinel-2 Imagery | en |
| dc.type | Article | en |