Research on sea island region recognition on improved attention U-Net

In the context of the worldwide exploration of islands, monitoring changes in island shorelines and areas is of great significance to the ecological protection and economic development of islands. The island coastline's dynamic change directly impacts the stability of coastal areas, biodiversity and human activities. Therefore, effectively monitoring island coastlines and areas is important to global environmental monitoring. The traditional way of island shoreline monitoring mainly relies on manual field measurements and semi-automatic processing of remote sensing images. With the progress of deep learning technology, semantic segmentation has become an important technique for remote sensing image classification. In this paper, the improved Attention U-Net model is applied to the recognition task of island shoreline and compared with several other deep learning models of U-Net. Each model has its characteristics in terms of structure and performance, providing different perspectives and solutions for island shoreline recognition. This paper comprehensively describes the architecture of the improved Attention U-Net model and the effectiveness of different models in recognising sea island areas.