Flood detection and monitoring are essential components of disaster management; however, traditional methods relying on remote sensing or IoT-based solutions are often cost-prohibitive and complex to deploy. This study proposes a deep learning-based flood monitoring system utilizing the YOLOv8-OBB model for real-time and accurate water level detection. The system was implemented near Irwin Bridge, Sangli, where NVR cameras continuously capture footage of a painted scale on the river wall. A custom dataset was created, and several deep learning models, including CNN (TensorFlow) and multiple YOLO variants, were trained and evaluated. Experimental results show that YOLOv8-OBB outperforms other models in terms of accuracy and robustness for water level estimation. The system triggers alerts when water levels exceed critical thresholds, enabling timely flood warnings. Supported by SMKC Sangli, this research demonstrates a cost-effective, scalable, and precise approach to flood monitoring using deep learning techniques.