Environmental interference and the trade-off between accuracy and efficiency hinder offshore wind turbine (OWT) identification in complex shallow seas. To address this, we propose GCRC, an innovative two-stage framework that integrates a gray-level co-occurrence matrix (GLCM)-enhanced collaborative random forest (RF) with a convolutional neural network for OWT identification under complex marine environments. By innovatively employing RF as an efficient region proposal network (RPN) to filter complex backgrounds prior to the lightweight ResNet18 inference, this method drastically reduces computational redundancy. Validated using 2024 Sentinel-2 imagery of China, GCRC achieved a precision of 0.9937 and an F1-score of 0.9949. Leveraging Google Earth Engine (GEE), nationwide inference was completed in less than 0.5 h on a consumer-grade GPU, representing a substantial reduction compared to the 20+ h required by existing models, while successfully identifying 7578 OWTs. Spatial analysis reveals distinct regional patterns: northern regions focus on shallow waters, while southern regions expand into deeper environments. This framework establishes a scalable, efficient paradigm for large-scale marine monitoring, offering robust support for the Digital Earth vision.
Jin et al. (Tue,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: