比较研究：基于深度学习的合成孔径雷达图像冰川崩解前缘描绘
计算机视觉与模式识别
2026-04-20 v2 机器学习
摘要
持续监测冰川崩解前缘对于海平面上升预测至关重要。本研究对深度学习系统在合成孔径雷达图像中描绘前缘的性能进行了基准测试。虽然深度学习系统的误差高达221米，但人工标注者的偏差仅为38米，凸显了进一步研究的必要性。
引用
@article{arxiv.2501.05281,
title = {Comparison Study: Glacier Calving Front Delineation in Synthetic Aperture Radar Images With Deep Learning},
author = {Nora Gourmelon and Konrad Heidler and Erik Loebel and Daniel Cheng and Julian Klink and Anda Dong and Fei Wu and Noah Maul and Moritz Koch and Marcel Dreier and Dakota Pyles and Thorsten Seehaus and Matthias Braun and Andreas Maier and Vincent Christlein},
journal= {arXiv preprint arXiv:2501.05281},
year = {2026}
}
备注
Accepted as short paper in IEEE Transactions on Pattern Analysis and Machine Intelligence