Rainy Image Segmentation, Flood Monitoring, and Confidence-Aware Civic Data

Abstract

Rainy image segmentation is essential for flood monitoring because weather degradation often occurs during periods of high urban risk. Domain adaptation under rainy conditions improves the transferability of segmentation models across changing visual environments, while continual rain removal supports long-term monitoring in outdoor systems. LiDAR-guided hyperspectral fusion contributes multimodal sensing methods that can strengthen classification under complex environmental conditions. Flood relocation research provides the disaster-risk context for interpreting visual evidence, while urban sub-center studies identify where monitoring may be most important. Digital civic data from Web3 platforms adds another layer of social and political behavior that may affect governance responses. LLM confidence research is relevant when automated systems communicate visual or policy evidence to decision-makers. This literature cluster supports flood monitoring systems that combine weather-robust vision, multimodal fusion, civic data, and confidence-aware decision support.