TrainSim

TrainSim是一个用于生成LiDAR和相机数据集的铁路模拟框架，由意大利比萨的圣安娜高等学校的卓越机器人与人工智能学院和热那亚的日立铁路STS共同开发。该数据集旨在为深度神经网络训练和测试提供大量标记数据，适用于视觉里程计、LiDAR里程计、图像分割、点云分割、图像深度估计和惯性导航等任务。数据集内容包括RGB图像、深度图像、分割图像、点云、分割点云以及车辆姿态、速度和加速度数据。此外，TrainSim还支持生成任意大的标记数据集，并通过模拟传感器（如LiDAR、相机和IMU）产生与物理传感器相似的数据。

TrainSim is a railway simulation framework for generating LiDAR and camera datasets, jointly developed by the School of Excellence in Robotics and Artificial Intelligence of the Sant'Anna School of Advanced Studies in Pisa, Italy, and Hitachi Rail STS in Genoa. This dataset is designed to provide large-scale annotated data for training and testing deep neural networks, and is applicable to tasks including visual odometry, LiDAR odometry, image segmentation, point cloud segmentation, image depth estimation, and inertial navigation. The dataset includes RGB images, depth images, segmented images, point clouds, segmented point clouds, as well as vehicle pose, velocity, and acceleration data. Furthermore, TrainSim supports the generation of arbitrarily large annotated datasets, and can produce data consistent with that of physical sensors by simulating sensing devices such as LiDAR, cameras, and IMUs.