Research on wave height and period observation method for nearshore waters based on YOLOv8 video image analysis

Coastal wave observation is crucial for marine hydrology, forecasting, disaster prevention and mitigation, and coastal zone management. This study introduces the application of the YOLOv8 model to accurately estimate wave height and period in nearshore waters. Unlike traditional methods such as buoys, pressure wave sensors, and acoustic wave sensors, we utilized high-definition video captured by a camera with a resolution of 2304×1296 to analyze buoy motion. By employing the YOLOv8 deep learning model for training and inference, we extracted image data reflecting buoy oscillations to compute wave height and period parameters. The model achieved pixel-level (centimeter-level) accuracy in buoy position recognition, aligning closely with manual observations from coastal wave sensors. Results demonstrate that using the YOLOv8 model to extract buoy position features from high-definition buoy videos for estimating wave height and period is feasible, offering an efficient and cost-effective approach for marine wave observation.