Deep Learning for Rapid Identification and Assessment of Disaster Areas Based on Satellite Images

Abstract

350 million people are affected annually by natural disasters, in addition to financial losses amounting to billions of dollars. When these disasters occur, obtaining correct information about damage locations leads to a rapid and effective response by rescue teams, thus saving the largest number of lives. Rescue teams rely on satellite images to assess the locations and severity of damage caused by disasters. However, rescue teams need to follow a specific approach that enables them to analyze huge amounts of satellite images accurately and quickly, which represents a major challenge for them. Deep learning can be used for rapid identification and assessment of disaster areas to overcome these challenges and provide assistance and support efforts. In this research, Siamese U-Net deep learning system with attention technique was applied on two groups of satellite images (pre- and post-disaster) for semantic segmentation of buildings as a first step. Then, the decoder extracts high-dimensional feature vectors for damage level classification. The Siamese networks detect changes in the input data and combat noise by focusing on learning relative differences. Self-attention modules were included to capture important information from the feature vectors, thus enabling the system to focus on the areas surrounding buildings. The proposed system was evaluated on xBD, a benchmark dataset for building damage assessment, and achieved the best results for segmentation (IoU = 88%) and classification (overall F1=80.2%). Many comparisons were made with related works, where the proposed system achieved high segmentation accuracy and more reliable classification results in both numerical and visual experiments. The proposed system excelled in providing accurate classification results despite the presence of noise (buildings covered by trees or clouds) in the input samples, and the difference in the image capture angle of the satellites, in addition to other challenges.