A Group-Wise Feature Enhancement-and-Fusion Network with Dual-Polarization Feature Enrichment for SAR Ship Detection

Ship detection in synthetic aperture radar (SAR) images is a significant and challenging task. However, most existing deep learning-based SAR ship approaches are confined to single-polarization fail leverage dual-polarization characteristics, which increases the difficulty of further improving performance. One problem that requires solution how make full use characteristics excavate polarization features using network. To tackle problem, we propose group-wise feature enhancement-and-fusion network with enrichment (GWFEF-Net) for better detection. GWFEF-Net offers four contributions: (1) (DFE) enriching library suppressing clutter interferences facilitate extraction; (2) enhancement (GFE) enhancing each semantic highlight region; (3) fusion (GFF) fusing multi-scale realize features’ information interaction; (4) hybrid pooling channel attention (HPCA) modeling balance feature’s contribution. We conduct sufficient ablation studies verify effectiveness Extensive experiments on Sentinel-1 dataset demonstrate superior performance GWFEF-Net, 94.18% average precision (AP), compared other ten competitive methods. Specifically, can yield 2.51% AP improvement second-best method.