Building Damage Mapping with Self-PositiveUnlabeled Learning

Abstract

Humanitarian organizations must have fast and reliable data to respond to disasters. Deep learning approaches are difficult to implement in real-world disasters because it might be challenging to collect ground truth data of the damage situation (training data) soon after the event. The implementation of recent self-paced positive-unlabeled learning (PU) is demonstrated in this work by successfully applying to building damage assessment with very limited labeled data and a large amount of unlabeled data. Self-PU learning is compared with the supervised baselines and traditional PU learning using different datasets collected from the 2011 Tohoku earthquake, the 2018 Palu tsunami, and the 2018 Hurricane Michael. By utilizing only a portion of labeled damaged samples, we show how models trained with self-PU techniques may achieve comparable performance as supervised learning.

7 pages, 1 figure, Artificial Intelligence for Humanitarian Assistance and Disaster Response Workshop, NeurIPS 2021

References in corpus (7)

• Building Damage Detection in Satellite Imagery Using Convolutional Neural Networks
• Self-PU: Self Boosted and Calibrated Positive-Unlabeled Training
• Assessing Post-Disaster Damage from Satellite Imagery using Semi-Supervised Learning Techniques
• Assessing out-of-domain generalization for robust building damage detection
• Multi-class segmentation under severe class imbalance: A case study in roof damage assessment
• An Attention-Based System for Damage Assessment Using Satellite Imagery
• Cross-directional Feature Fusion Network for Building Damage Assessment from Satellite Imagery