WaterScenes
收藏资源简介:
WaterScenes数据集是专为水面自动驾驶设计的,结合了4D雷达和单目相机技术,以实现全天候的目标信息识别,包括颜色、形状、纹理、范围、速度、方位和高度。数据集关注水面上的典型静态和动态物体,对相机图像进行像素级标注,对雷达点云进行点级标注。除了基本的感知任务,如物体检测、实例分割和语义分割,还提供了自由空间分割和水线分割的标注。通过多任务和多模态数据,数据集支持对雷达和相机单一模态以及融合模态的基准实验。实验结果表明,4D雷达-相机融合能显著提高水面感知的准确性和鲁棒性,尤其是在恶劣光照和天气条件下。
The WaterScenes dataset is specifically designed for autonomous driving on water surfaces. It integrates 4D radar and monocular camera technologies to enable all-weather target information recognition, covering attributes including color, shape, texture, range, speed, azimuth and height. The dataset focuses on typical static and dynamic objects on water surfaces, with pixel-level annotations provided for camera images and point-level annotations for radar point clouds. In addition to basic perception tasks such as object detection, instance segmentation and semantic segmentation, it also supplies annotations for free space segmentation and waterline segmentation. Leveraging multi-task and multi-modal data, the dataset supports benchmark experiments for single-modal (radar-only, camera-only) and multi-modal fusion-based approaches. Experimental results show that 4D radar-camera fusion can significantly improve the accuracy and robustness of water surface perception, especially under adverse lighting and weather conditions.

- 1WaterScenes: A Multi-Task 4D Radar-Camera Fusion Dataset and Benchmark for Autonomous Driving on Water Surfaces苏州西交利物浦大学高级技术学院 · 2023年



