Abstract: The rapid development of spaceborne Synthetic Aperture Radar (SAR) imaging technology has brought about massive remote sensing data, posing a serious challenge to onboard real-time processing capabilities. Traditional target detection algorithms struggle to meet the demands of onboard real-time detection due to computational resource limitations as well as issues such as SAR image noise and large variations in target scales. This paper proposes a real-time ship detection algorithm for spaceborne SAR based on dynamic convolution and multi-scale attention enhancement. Firstly, a lightweight feature extraction module is constructed by integrating dynamic convolution kernels and the Ghost module, adaptively adjusting convolution parameters to handle variations in target scale. Secondly, a lightweight coordinate attention feature fusion module, which employs a dual-path spatial attention mechanism to enhance the feature representation of small targets, is designed. Finally, an efficient detection head is introduced to reduce computational redundancy through a decoupled structure and partial convolution. Comparative experiments on the SSDD dataset demonstrate that the proposed algorithm achieves an AP of 98.86%, with a 45.8% reduction in the number of parameters compared to the baseline model. This high-precision real-time target detection method, tailored for resource-constrained spaceborne platforms, significantly improves the response efficiency of maritime situational awareness and emergency monitoring, offering substantial engineering application value.