Breaking the ice: video segmentation for close-range ice-covered waters

Téléchargement	Voir la version finale : Breaking the ice: video segmentation for close-range ice-covered waters (PDF, 7.9 Mio)
DOI	Trouver le DOI : https://doi.org/10.1016/j.oceaneng.2025.122105
Auteur	Rechercher : MacMillan, Corwin Grant Jeon Identifiant ORCID : https://orcid.org/0009-0008-0376-9137; Rechercher : Scott, K. Andrea; Rechercher : Garvin, Matthew¹Identifiant ORCID : https://orcid.org/0009-0006-7623-2563; Rechercher : Pan, Zhao Identifiant ORCID : https://orcid.org/0000-0003-1654-3205
Affiliation	Conseil national de recherches Canada. Génie océanique, côtier et fluvial
Format	Texte, Article
Sujet	ice classification; machine learning; video segmentation; optical flow; occlusion mitigation; lens occlusion
Résumé	Rapid ice recession in the Arctic Ocean, with predictions of ice-free summers by 2060, opens new maritime routes but requires reliable navigation solutions. Current approaches rely heavily on subjective expert judgment, underscoring the need for automated, data-driven solutions. This study leverages machine learning to assess ice conditions using ship-borne optical data, and introduces a semi-manual, region-based annotation technique to alleviate the high cost of manual annotation for close-range shipborne ice data. The resulting dataset was used to train and evaluate the proposed video segmentation model, UPerFlow, which advances the SegFlow architecture by incorporating a six-channel ResNet encoder, separate UPerNet-based segmentation decoders for each image, PWCNet as the optical flow encoder, and cross-connections that integrate bi-directional flow features without loss of latent information. The proposed architecture outperforms baseline image segmentation networks by an average 38 % in occluded regions, demonstrating the robustness of video segmentation in addressing challenging Arctic conditions.
Date de publication	2025-11-30
Date de création	2025-07-30
Maison d’édition	Elsevier BV
Licence	Creative Commons, Attribution - Pas d’utilisation commerciale 4.0 International (CC BY-NC 4.0) https://creativecommons.org/licenses/by-nc/4.0/deed.fr
Dans	Ocean Engineering 340, nº P2, 122105 (30 novembre 2025).
Langue	anglais
Publications évaluées par des pairs	Oui
Identificateur	S0029801825017895
Exporter la notice	Exporter en format RIS
Signaler une correction	Signaler une correction (s'ouvre dans un nouvel onglet)
Identificateur de l’enregistrement	71bb9aa8-3cb8-44e4-9a99-2922c4b8afc8
Enregistrement créé	2025-07-31
Enregistrement modifié	2025-08-01

Détails de la page

Par :

Conseil national de recherches Canada

Date de modification :: 2026-04-18