Aquaculture Areas Extraction Model Using Semantic Segmentation from Remote Sensing Images at the Maowei Sea of Beibu Gulf

The extraction of aquaculture areas from high-resolution remote sensing images is crucial for effective coastal management and resource preservation. This study introduces SwinNet, a semantic segmentation model leveraging multi-scale feature fusion to enhance the extraction of aquaculture areas, particularly in the Maowei Sea of the Beibu Gulf, China. Utilizing the Swin Transformer backbone and a novel Parallel Pooling Attention Module (PPAM), SwinNet minimizes background noise and improves segmentation accuracy. SwinNet achieved a pixel accuracy of 96.53% and an intersection over the union of 93.07% on an aquaculture dataset, demonstrating superior performance in overcoming noise and accurately extracting aquaculture areas. SwinNet offers an effective solution for large-scale, high-precision monitoring of coastal aquaculture, with potential broader applicability in aquatic resource conservation and management.