With the growing applications of water operations, water surface object detection tasks are facing new challenges. In this paper, we focus on improving the performance of water surface small object detection. Due to the limitations of single-sensor in water environments, we propose RCFNet, a novel small object detection method based on radar-vision fusion. RCFNet fuses features captured by radar and camera in multiple stages to generate more effective target feature representations for small object detection on water surfaces. In particular, we propose a multi-frame radar feature fusion module and an image-guided radar feature enhancement module to enhance the radar features. This method fully utilizes radar and camera data information, improving the performance of small object detection on water surfaces. RCFNet was evaluated on the publicly released water surface floating dataset Flow, achieving an Average Precision (AP50) of 0.9316, which is a state-of-the-art result.