ViT-Large with focal loss on Sentinel-1 SAR data reaches 69.6% held-out accuracy and 83.9% precision on minority multi-year ice, establishing a data-centric baseline for sea ice classification.
A comparative study of data input selection for deep learning-based automated sea ice mapping
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
A Data-Centric Vision Transformer Baseline for SAR Sea Ice Classification
ViT-Large with focal loss on Sentinel-1 SAR data reaches 69.6% held-out accuracy and 83.9% precision on minority multi-year ice, establishing a data-centric baseline for sea ice classification.