Building Damage Detection in Satellite Imagery Using Convolutional Neural Networks

Abstract

In all types of disasters, from earthquakes to armed conflicts, aid workers need accurate and timely data such as damage to buildings and population displacement to mount an effective response. Remote sensing provides this data at an unprecedented scale, but extracting operationalizable information from satellite images is slow and labor-intensive. In this work, we use machine learning to automate the detection of building damage in satellite imagery. We compare the performance of four different convolutional neural network models in detecting damaged buildings in the 2010 Haiti earthquake. We also quantify how well the models will generalize to future disasters by training and testing models on different disaster events.

References in corpus (1)

• Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks

Cited by in corpus (13)

• BDANet: Multiscale Convolutional Neural Network with Cross-directional Attention for Building Damage Assessment from Satellite Images
• Building Disaster Damage Assessment in Satellite Imagery with Multi-Temporal Fusion
• Semantic-aware Dense Representation Learning for Remote Sensing Image Change Detection
• Continuous Cross-resolution Remote Sensing Image Change Detection
• Assessing Post-Disaster Damage from Satellite Imagery using Semi-Supervised Learning Techniques
• Towards Cross-Disaster Building Damage Assessment with Graph Convolutional Networks
• An Attention-Based System for Damage Assessment Using Satellite Imagery
• Cross-directional Feature Fusion Network for Building Damage Assessment from Satellite Imagery
• Multi-class segmentation under severe class imbalance: A case study in roof damage assessment
• Cross Branch Feature Fusion Decoder for Consistency Regularization-based Semi-Supervised Change Detection
• Disaster mapping from satellites: damage detection with crowdsourced point labels
• Leveraging Domain Adaptation for Low-Resource Geospatial Machine Learning
• Building Damage Mapping with Self-PositiveUnlabeled Learning