FMARS: Annotating Remote Sensing Images for Disaster Management Using Foundation Models

Very-High Resolution (VHR) remote sensing imagery is increasingly accessible, but often lacks annotations for effective machine learning applications. Recent foundation models like GroundingDINO [1] and Segment Anything (SAM) [2] provide opportunities to automatically generate annotations. This study introduces FMARS (Foundation Model Annotations in Remote Sensing), a methodology leveraging VHR imagery and foundation models for fast and robust annotation. We focus on disaster management and provide a large-scale dataset with labels obtained from pre-event imagery over 19 disaster events, derived from the Maxar Open Data initiative. We train segmentation models on the generated labels, using Unsupervised Domain Adaptation (UDA) techniques to increase transferability to real-world scenarios. Our results demonstrate the effectiveness of leveraging foundation models to automatically annotate remote sensing data at scale, enabling robust downstream models for critical applications. Code and dataset are available at https://github.com/links-ads/igarss-fmars.

1. INTRODUCTION

Remote Sensing (RS), and especially Very-High Resolution (VHR) images, represent a crucial resource for many real-world scenarios, including land use and land cover monitoring, urban planning, and disaster management. Recent advances in satellite technologies have allowed for increasingly accessible remote sensing data, also thanks to public and private programs such as the European Union’s Copernicus program [3] or the Maxar Open Data program [4], which helps to democratize the access to medium and high-resolution data for research purposes on a global scale. However, using such data with supervised machine learning mighe be challenging due to the limited availability of high-quality annotations.