KITTI-CARLA
收藏资源简介:
KITTI-CARLA数据集是由巴黎矿业技术学院的Jean-Emmanuel Deschaud基于CARLA模拟器创建的,旨在模拟KITTI数据集的环境和传感器配置。该数据集包含7个序列,每个序列5000帧,共计35000帧,涵盖城市、郊区、山区、乡村和高速公路等多种环境。数据集通过模拟车载的Velodyne HDL64 LiDAR和两个彩色相机生成,用于测试和比较语义分割、激光雷达和/或图像里程计的方法。数据集的创建过程中,使用了CARLA模拟器的高频率更新(1000Hz)来精确模拟传感器的动态效果。该数据集主要应用于自动驾驶领域,旨在通过合成数据提升真实数据上的转移学习方法。
The KITTI-CARLA dataset was developed by Jean-Emmanuel Deschaud from the Paris Institute of Mining and Technology, based on the CARLA simulator. It is designed to replicate the environment and sensor configurations of the KITTI dataset. This dataset includes 7 sequences, each containing 5000 frames, totaling 35,000 frames, covering diverse environments such as urban, suburban, mountainous, rural, and highway scenarios. Generated by simulating the on-board Velodyne HDL64 LiDAR and two color cameras, the dataset is used for testing and comparing methods for semantic segmentation, LiDAR odometry and/or visual odometry. During its creation, the high-frequency update (1000Hz) of the CARLA simulator was utilized to precisely simulate the dynamic effects of the sensors. This dataset is primarily applied in the field of autonomous driving, aiming to enhance transfer learning methods for real-world data via synthetic data.




