Citation

Tan, Hui Hui and Lim, Sin Liang and Jatmiko, Wisnu and Azizah, Kurniawati and Hilman, Muhammad Hafizhuddin (2026) A Review of Deep Learning Methodologies for Building Damage Assessment Using Optical Satellite Imagery. IEEE Access, 14. pp. 93989-94013. ISSN 2169-3536

Text 8.pdf - Published Version Restricted to Repository staff only Download (3MB)

Abstract

Catastrophic events such as earthquakes, hurricanes, floods, and wildfires can damage large numbers of buildings within a short time, creating an urgent need for rapid and coordinated disaster response. One of the major challenge of this is obtaining reliable damage information quickly, because ground inspections and manual interpretation of satellite images are slow, hazardous, and labor-intensive. Remote sensing enables wide-area observation through satellite and aerial imagery, while recent advances in artificial intelligence, especially deep learning-based computer vision, have improved the automation of post-disaster building damage assessment. These models can support faster and more scalable interpretation of optical imagery, although their performance depends on data quality, disaster type, class balance, and deployment conditions. This study reviews deep learning methodologies for building damage assessment using optical satellite imagery. The main contributions are: 1) a review of the historical progression from manual inspection to computer vision-based assessment; 2) a categorization of real-world and synthetic datasets used for benchmarking; 3) an organization of recent optical deep learning approaches according to temporal input strategy and model architecture; 4) critical synthesis of major methodological patterns, including model architectures, temporal input strategies, class imbalance, and generalizability; and 5) a discussion of evaluation practices, deployment limitations, ethical issues, and future directions such as learning with fewer manual labels and Vision-Language Models (VLMs).

Item Type:	Article
Uncontrolled Keywords:	Building damage assessment, deep learning, optical satellite imagery, remote sensing, disaster response
Subjects:	T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK7800-8360 Electronics > TK7871 Electronics--Materials
Divisions:	Faculty of Artificial Intelligence & Engineering (FAIE)
Depositing User:	Ms Suzilawati Abu Samah
Date Deposited:	31 Jul 2026 06:10
Last Modified:	31 Jul 2026 06:10
URII:	http://shdl.mmu.edu.my/id/eprint/16408

Downloads

Edit (login required)