This study tested smart web cameras to measure water levels during coastal flooding. The web cameras were placed in 10 different locations across the U.S. and took pictures every six minutes for three months. The cameras used 4k high-resolution optical sensors and infrared night-vision to capture clear images—even in the dark or during bad weather. An Xception machine learning computer model, trained using a week’s worth of these images, learned how to identify and measure water levels from prior pictures captured at each site-specific sensor location. It used artificial intelligence to recognize changes in the water surface and correct for issues like fog or rain. The system proved to be highly accurate—able to measure most water levels within just a few centimeters of the true value. Results were cross referenced using USGS radar water monitoring gages. This approach could help improve how we monitor flooding in real time using low-cost, camera-based systems.