Flood Segmentation on Sentinel-1 SAR Imagery with Semi-Supervised Learning
Flood Segmentation on Sentinel-1 SAR Imagery with Semi-Supervised Learning
复制标题
使用半监督学习对 Sentinel-1 SAR 图像进行洪水分割
DOI：
--
复制
发表时间：
2021
期刊：
影响因子：
--
通讯作者：
Siddha Ganju
中科院分区：
文献类型：
--
作者：
Sayak Paul;Siddha Ganju
洪水在世界各地肆虐，造成数十亿美元的损失，并摧毁社区、生态系统和经济。NASA的Impact Flood Detection竞赛要求参与者在监督环境中使用合成孔径雷达（SAR）图像进行训练后预测淹没像素。我们提出了一个半监督学习伪标记方案，该方案从U-Net集成中获得置信度估计，逐步提高准确性。具体来说，我们使用一种循环方法，涉及多个阶段：（1）使用提供的高置信度手工标记数据训练多个U-Net架构的集成模型，并在整个未标记测试数据集上生成伪标签或低置信度标签，然后（2）过滤出质量生成的标签，（3）将生成的标签与先前可用的高置信度手工标记数据集联合收割机组合。该同化的数据集用于下一轮训练集合模型，并且重复循环过程，直到性能改善达到平台。我们使用条件随机场对结果进行后处理。我们的方法在Sentinel-1数据集上设置了一个新的最先进的状态，具有0.7654 IoU，比0.60 IoU基线有了令人印象深刻的改进。我们的方法与所有代码和模型一起发布，也可以用作Sentinel-1数据集的开放科学基准。
Floods wreak havoc throughout the world, causing billions of dollars in damages, and uprooting communities, ecosystems and economies. The NASA Impact Flood Detection competition tasked participants with predicting flooded pixels after training with synthetic aperture radar (SAR) images in a supervised setting. We propose a semi-supervised learning pseudo-labeling scheme that derives confidence estimates from U-Net ensembles, progressively improving accuracy. Concretely, we use a cyclical approach involving multiple stages (1) training an ensemble model of multiple U-Net architectures with the provided high confidence hand-labeled data and, generated pseudo labels or low confidence labels on the entire unlabeled test dataset, and then, (2) filter out quality generated labels and, (3) combine the generated labels with the previously available high confidence hand-labeled dataset. This assimilated dataset is used for the next round of training ensemble models and the cyclical process is repeated until the performance improvement plateaus. We post process our results with Conditional Random Fields. Our approach sets a new state-of-the-art on the Sentinel-1 dataset with 0.7654 IoU, an impressive improvement over the 0.60 IoU baseline. Our method, which we release with all the code and models, can also be used as an open science benchmark for the Sentinel-1 dataset.
影响因子：
14.9
作者：
Arnab, Anurag;Zheng, Shuai;Torr, Philip H. S.
通讯作者：
Torr, Philip H. S.