"Lidar_Ship_detectedBY_MID360"

"This 3D point cloud dataset is captured using a Livox Mid-360 LiDAR sensor, focusing on real-scene vessel targets on inland lakes, and is fully structured to be directly compatible with the OpenPCDet framework for 3D object detection tasks. It mainly consists of small fishing boats under natural water environments, with rich point cloud features including hull structures, shipboard details, and partial water surface reflections, covering diverse real-world lake navigation scenarios with variations in boat sizes, viewing angles, and environmental conditions. The dataset is organized in a standard OpenPCDet-compatible structure, with core directories including points\/ for raw point cloud data, labels\/ for manually annotated 3D bounding box labels, ImageSets\/ for train\/val split files, and gt_database\/ for ground-truth point cloud patches optimized for data augmentation. Pre-generated pickle files (custom_dbinfos_train.pkl, custom_infos_train.pkl, custom_infos_val.pkl) store dataset metadata and sample information, eliminating the need for additional preprocessing. The data provides accurate, dense, and scene-consistent point cloud samples, supporting research on 3D object detection, point cloud segmentation, and maritime target perception for unmanned surface vehicles (USVs) and marine engineering applications."