Region-level labels in ice charts can produce pixel-level segmentation for Sea Ice types

TL;DR

This paper presents a weakly supervised deep learning method that uses regional labels from ice charts to achieve high-resolution pixel-level sea ice classification, outperforming fully supervised benchmarks.

Contribution

It introduces a novel regional loss approach that leverages lower-resolution labels to improve pixel-level segmentation accuracy in sea ice mapping.

Findings

Outperforms fully supervised U-Net benchmark

Achieves higher mapping resolution and class accuracy

Demonstrates effectiveness on AI4Arctic Sea Ice Challenge Dataset

Abstract

Fully supervised deep learning approaches have demonstrated impressive accuracy in sea ice classification, but their dependence on high-resolution labels presents a significant challenge due to the difficulty of obtaining such data. In response, our weakly supervised learning method provides a compelling alternative by utilizing lower-resolution regional labels from expert-annotated ice charts. This approach achieves exceptional pixel-level classification performance by introducing regional loss representations during training to measure the disparity between predicted and ice chart-derived sea ice type distributions. Leveraging the AI4Arctic Sea Ice Challenge Dataset, our method outperforms the fully supervised U-Net benchmark, the top solution of the AutoIce challenge, in both mapping resolution and class-wise accuracy, marking a significant advancement in automated operational sea ice mapping.