Unmanned surface vehicles (USVs) are increasingly used in bathymetric survey, maritime surveillance, and maintenance applications. However, detecting obstacles in the maritime environment poses significant challenges due to sea clutter, background variability, and other factors. Although various sensors such as radar, AIS, LiDAR, and camera have been used for obstacle detection, they have limitations in terms of range and effectiveness, especially at different USV speeds. In this paper, we present a real-time obstacle detection system for USVs in the maritime environment by explaining the used sensors, the architecture and the implemented algorithms. We focus on evaluating the performance of the YOLOV7 network, which forms the basis of our fusion algorithm. Our results demonstrate that our system can effectively detect maritime objects in real-time, providing improved safety and efficiency for USV operations. Additionally, we provide an open solution for visualizing the detected targets on a chart plotter navigation.