Abstract: In order to obtain the spatial distribution and dynamic change of Marine aquaculture quickly, accurately and in a large range, this study designed a remote sensing dynamic monitoring system based on deep learning. Using C# and Microsoft Visual Studio 2012, this study developed an integrated platform incorporating ArcGIS Engine, DevExpress components, and a deep learning-based aquaculture zone extraction model. The platform combines remote sensing data preprocessing, intelligent seawater aquaculture extraction, spatial analysis, and thematic mapping. This paper details the system's functional requirements, module design, development environment, and implementation features. Using GF-2 remote sensing imagery, the MSU-ResUnet model was applied to rapidly and accurately map the distribution of raft and cage aquaculture zones in Zhao'an Bay, Fujian Province, in 2020, followed by a compliance analysis for planning. The system achieves rapid and high-precision extraction and dynamic analysis of large-scale marine aquaculture zones, significantly enhancing automation and efficiency in aquaculture information extraction. It provides scientific and intuitive decision support for coastal aquaculture planning management and environmental protection.