Abstract: Due to the presence of a large amount of background clutter and the diversity of targets on the sea surface, sea-surface small target detection has always been a challenging problem. In recent years, sea surface small target detection based on deep learning is gradually becoming popular and has good detection performance. However, such methods heavily rely on the quantity and quality of training samples. Typically, the number of target samples is much smaller than the number of sea clutter samples, which can bring difficulties to training and reduce detection performance. Based on this, this paper proposes a sea surface small target detection method based on target data augmentation. By analyzing the time-frequency characteristics of the target, similar feature distribution samples are generated based on the Generative adversarial network (GAN), so as to expand the target data set, narrow the number gap between target samples and sea clutter samples, and make more target information be used in the training process of the detector to improve the performance of the detector. The experimental results on the IPIX dataset indicate that this method can achieve better detection performance. The experimental results also indicate that the proposed detector can achieve control of false alarm rate.