Design of Intelligent Detection System for Flash Flood Disasters by Integrating YOLOv8 and Remote Sensing Images

This paper proposes an intelligent detection system for flash flood disasters that integrates YOLOv8 and remote sensing images. It makes full use of the large-scale and multi-temporal characteristics of remote sensing data, and combines the efficiency and accuracy advantages of the YOLOv8 target detection algorithm to achieve rapid identification and risk classification of flooded areas. The system design covers key links such as remote sensing image preprocessing, elevation data fusion, model customization training, multi-temporal change analysis and risk visualization output. Experiments based on the Sen1Floods11 dataset show that the system achieves high-precision detection with an average F1-score of 0.90 in typical flash flood scenarios, and the inference delay is controlled within 80 milliseconds, which is significantly better than traditional methods such as NDWI, SVM and FCN. The research results verified the comprehensive advantages of the constructed system in terms of accuracy, efficiency and practical deployment, and provided technical support and engineering application basis for intelligent early warning and emergency response of mountain floods.