In order to improve the detection accuracy of multi-scale remote sensing ship targets in complex scenes, a feature enhancement single shot multi-scale detector is proposed. Firstly, the shallow feature enhancement module is designed to improve the feature extraction ability of the shallow network in the pyramid structure of Single Shot MultiBox Detector(SSD). Then the deep feature fusion module is designed to replace the deep network in the pyramid structure of SSD to improve the feature extraction ability of deep network. Finally, the image features are matched with candidate frames of different aspect ratios to adapt to remote sensing image targets of different scales. The experiments tested on the optical remote sensing image dataset demonstrate that the proposed method can adapt to target detection under different background and effectively improve the detection performance of multi-scale remote sensing targets in complex scenes. On the extended experiment, the proposed method performance over SSD in blurry target detection.