Abstract: Compact High-Frequency Surface Wave Radar (HFSWR) with low transmit power and small aperture array leads to an increase of weak targets in the radar echoes, which in turn degrades the performance of range-Doppler spectrum-based target detection method, and weakens the target detection capability. In order to improve the detection performance of compact HFSWR on weak target signals, this paper proposes a weak target detection method for HFSWR based on Synchroextracting Transform-Convolutional Neural Network (SET-CNN). Firstly, the Signal-to-Noise Ratio (SNR) method is used to suppress the sea clutter in the signal and reduce the influence of the clutter time-frequency ridges on the target detection. Then the time-frequency ridge sample database is conducted based on the SET time-frequency spectrum, and then the time-frequency ridges are classified through the CNN. Thereafter the vessel targets are detected by the post-processing of the classification results. The proposed target detection method is validated by simulation and measured data, and the results show that the method proposed can effectively detect weak targets and improve the up-to-target detection performance of compact high-frequency surface wave radar.