ADS Capstone Chronicles Revised

19

6.3 Recommended Future Studies

Future studies should focus on enhancing the scalability and accuracy of ML models by incorporating diverse datasets from various disaster types, including earthquakes, wildfires, and tsunamis. Research should explore the integration of additional data sources, such as social media and real-time sensor data, to provide more comprehensive situational awareness during and after disaster events. Additionally, the studies should explore the integration of change detection techniques to analyze differences between pre-and post-disaster images, enabling a more precise understanding of structural damage. While this study used MobileNet and ResNet-50, other advanced models such as DenseNet, EfficientNet, and vision transformers could be investigated for their potential to capture complex features and subtle differences between damage categories. Studies could also investigate the use of advanced techniques, such as deep reinforcement learning, to optimize resource allocation and decision making in dynamic disaster environments. Further research into the application of real-time damage assessment models in post-disaster recovery could help refine the tools for immediate deployment. Additionally, exploring the impact of higher resolution satellite imagery and multi-spectral data could improve the precision of damage classification and enhance model robustness. Finally, future studies could examine the ethical implications of using satellite imagery and ML in disaster response, focusing on privacy concerns and the equitable distribution of aid and infrastructure in affluent and impoverished cities, states, and nations.

285