Assessing Post-Disaster Damage from Satellite Imagery using Semi-Supervised Learning Techniques

TL;DR

This paper demonstrates that semi-supervised learning techniques can effectively assess disaster damage from satellite imagery with minimal labeled data, matching fully supervised models' performance across various disaster scenarios.

Contribution

The study applies and compares state-of-the-art semi-supervised learning methods to disaster damage assessment, showing they can achieve comparable results with significantly less labeled data.

Findings

SSL methods reach full supervised performance with less labeled data

MixMatch and FixMatch outperform supervised baseline in experiments

Models generalize well across different disaster types

Abstract

To respond to disasters such as earthquakes, wildfires, and armed conflicts, humanitarian organizations require accurate and timely data in the form of damage assessments, which indicate what buildings and population centers have been most affected. Recent research combines machine learning with remote sensing to automatically extract such information from satellite imagery, reducing manual labor and turn-around time. A major impediment to using machine learning methods in real disaster response scenarios is the difficulty of obtaining a sufficient amount of labeled data to train a model for an unfolding disaster. This paper shows a novel application of semi-supervised learning (SSL) to train models for damage assessment with a minimal amount of labeled data and large amount of unlabeled data. We compare the performance of state-of-the-art SSL methods, including MixMatch and FixMatch, to a supervised baseline for the 2010 Haiti earthquake, 2017 Santa Rosa wildfire, and 2016 armed conflict in Syria. We show how models trained with SSL methods can reach fully supervised performance despite using only a fraction of labeled data and identify areas for further improvements.