Abstract: Objectives To address the issues of strong noise and target scale variation in Synthetic Aperture Radar (SAR) imagery, this paper proposes an improved YOLOv8n model, SSD-YOLO, to enhance ship detection performance in complex marine scenarios. Methods Based on YOLOv8n, the model integrates three innovative modules. The SAR_SPPF module utilizes Ghost Convolution and dual-path pooling for efficient noise resistance. The C2f_SimAM module embeds a parameter-free SimAM attention mechanism to strengthen target responses. The C2f_DSConv module employs direction-sensitive depthwise separable convolutions to capture fine ship textures and directional information.Results On the SSDD dataset, the model achieved 97.0% precision, 96.4% recall, 99.0% mAP50, and 74.3% mAP50-95. Experiments on generalization were also conducted on the HRSID dataset.With approximately 3M parameters and 7.7G FLOPs, the model remains lightweight. Ablation studies confirmed the effectiveness of each module. Conclusions This research provides an efficient and robust solution for SAR ship detection by combining lightweight anti-noise, target enhancement, and direction-sensitive feature capture strategies.