“Related research has made new progress in the field of target extraction in synthetic aperture radar (SAR) images. Researchers have proposed a Transformer network S3T Net based on frequency domain spatial collaboration. This network combines frequency domain encoding units and visual Transformer encoding units, and performs well in fine extraction of typical targets in SAR images through collaborative weighted feature fusion and recursive frequency spatial refinement modules. Experimental results show that it surpasses the current optimal performance model on multiple public datasets, providing new methods and theoretical support for capturing ground information in high interference environments.”