A-BFPN: An Attention-Guided Balanced Feature Pyramid Network for SAR Ship Detection

Thanks to the excellent feature representation capabilities of neural networks, target detection methods based on deep learning are now widely applied in synthetic aperture radar (SAR) ship detection. However, multi-scale variation, small targets with complex background such as islands, sea clutter, and inland facilities SAR images increase difficulty for To performance, this paper, a novel network detection, termed attention-guided balanced pyramid (A-BFPN), is proposed better exploit semantic multilevel complementary features, which consists following two main steps. First, order reduce interferences from backgrounds, enhanced refinement module (ERM) developed enable BFPN learn dependency features channel space dimensions, respectively, enhances objects. Second, fusion (CAFN) model designed obtain optimized serious aliasing effects hybrid maps. Finally, we illustrate effectiveness method, adopting existing Ship Detection Dataset (SSDD) Large-Scale Dataset-v1.0 (LS-SSDD-v1.0). Experimental results show that method superior algorithms, especially under background.