ISSLIDE: InSAR dataset for Slow SLIding area DEtection with machine learning

Slow moving motions are mostly tackled by using the phase information of Synthetic Aperture Radar (SAR) images through Interferometric SAR (InSAR) approaches based on machine and deep learning. Nevertheless, to the best of our knowledge, there is no dataset adapted to machine learning approaches and targeting slow ground motion detections. With this dataset, we propose a new InSAR dataset for Slow SLIding areas DEtections (ISSLIDE) with machine learning. The dataset is composed of standardly processed interferograms and manual annotations created following geomorphologist strategies. A total of 200 independent moves are delineated on each of the 54 orthorectified interferograms of 2018 Sentinel-1 images over the French Alps. Both the coherence and phase difference are available with their respective shapefiles in the Raw Dataset. A second Ready to Be Used Dataset is also released and consists in 13,230 patches of interferograms (coherence and phase) surrounding the move. This dataset is provided for future user to develop new machine and deep learning methodologies in detecting and localizing slow moving areas with standard, multi-temporal or multi-interferogram studies without any additional process.

缓慢移动的物体主要通过对合成孔径雷达（SAR）图像的相位信息进行干涉合成孔径雷达（InSAR）处理来解决。然而，据我们所知，目前尚无适用于机器学习方法的、针对缓慢地面运动检测的专用数据集。本数据集旨在提出一个新的InSAR数据集，用于慢速滑动区域检测（ISSLIDE），并应用于机器学习。该数据集由标准处理的干涉图和依据地貌学家策略制作的人工标注组成。在2018年法国阿尔卑斯山的54张正射校正干涉图中，共描绘了200次独立移动。原始数据集中同时提供了相关形状文件中的相干性和相位差。此外，还发布了一个可立即使用的第二个数据集，由13,230个围绕移动区域的干涉图（相干性和相位）补丁组成。此数据集旨在为未来的用户开发新的机器和深度学习方法，以实现对缓慢移动区域的检测和定位，这些方法可以通过标准、多时相或多干涉图研究实现，而无需任何额外的处理。