<p>Flood disasters pose significant threats to human lives and infrastructure, necessitating advanced methods for the timely and accurate monitoring of water levels in rivers. This study introduces an innovative approach utilizing the recent YOLOv9 (You Only Look Once, version 9) deep learning model to detect and monitor river water flow levels effectively. By leveraging YOLOv9’s robust object detection capabilities, the proposed system achieves precise segmentation and real-time analysis of river water regions. The method not only identifies river water areas with high accuracy but also quantifies flow levels, providing critical data for flood prediction and management. Comprehensive testing was conducted using diverse datasets representing various river conditions and flow scenarios. The results demonstrate that YOLOv9 significantly outperforms traditional methods in terms of detection speed and accuracy, making it a valuable tool for enhancing flood disaster response and management strategies. This work contributes to the development of more resilient and proactive flood disaster management systems through the integration of state-of-the-art machine learning techniques, offering significant improvements in predictive capabilities and response effectiveness. The proposed system was trained and evaluated using real-time river flood datasets, achieving a notable accuracy rate of 98%, a mean average precision (mAP) of 97.5%, precision of 98%, and recall of 98% in predicting flood level characteristics.</p>