A Multi-Task Water Surface Visual Perception Network for Unmanned Surface Vehicles
Xu Ling Hu, Ziteng Geng, Ju He, Yuxin Shi, Yu Yang, Xiaomin Zhang · 2023
In recent years, the significance of unmanned surface vehicles (USVs) has grown substantially across a wide range of applications. The monocular camera, as the most common perception sensor deployed in USV, offers the inherent advantages of cost and can also provide rich semantic information. To harness the perceptual capabilities of a monocular camera for USVs, we propose a multitask water surface visual perception network which can accomplish various perception tasks, including boat detection, drivable area segmentation, and camera quality classification in water surface scenes. Moreover, the proposed method demonstrates significant accuracy and high interface speed compared to baseline methods and can be effectively deployed to low-power embedded platforms in USVs.