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

TL;DR

This paper presents FMARS, a methodology that uses foundation models to automatically annotate high-resolution remote sensing images for disaster management, enabling scalable and effective training of segmentation models.

Contribution

Introducing FMARS, a novel approach that leverages foundation models and domain adaptation to generate large-scale annotations for remote sensing disaster datasets.

Findings

Effective automatic annotation of VHR imagery using foundation models

Improved transferability of segmentation models via Unsupervised Domain Adaptation

Large-scale annotated dataset for disaster management applications

Abstract

Very-High Resolution (VHR) remote sensing imagery is increasingly accessible, but often lacks annotations for effective machine learning applications. Recent foundation models like GroundingDINO and Segment Anything (SAM) 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 \url{https://github.com/links-ads/igarss-fmars}.