SemanticRail3D: A 3D Point Cloud dataset with semantic annotations of railway environments

The SemanticRail3D dataset consists of a total of 438 .laz point clouds of railway environment. Each point cloud covers a track length of approximately 200 metres. In total, the dataset has approximately 2800 million points. The dataset labels contains a total of 11 classes, and also includes the track position of each railway line, as well as instance segmentation. See Readme file for more information.

搜集汇总

背景与挑战

背景概述

SemanticRail3D是一个专注于铁路环境的3D点云数据集，包含438个点云文件，总点数约2800亿，每个点云覆盖约200米轨道长度。数据集提供详细的语义标注，包括11个物体类别、轨道位置信息以及实例分割，适用于铁路场景的计算机视觉和机器学习研究。