SynthItLikeKITTI
收藏资源简介:
SynthItLikeKITTI数据集是由德国埃尔朗根-纽伦堡弗里德里希-亚历山大大学视觉计算系和e:fs TechHub GmbH合作创建的合成数据集。该数据集利用CARLA模拟器生成,通过领域随机化策略增强了数据集的多样性,并针对LiDAR点云的三维对象检测任务进行了优化。数据集包含了多个地图生成的样本,模仿了KITTI数据集的配置,可用于Offline使用,并提供了数据处理脚本以及最终的 LiDAR 变体数据,以方便未来研究和实验变种。
SynthItLikeKITTI is a synthetic dataset jointly developed by the Department of Visual Computing, Friedrich-Alexander University Erlangen-Nuremberg, Germany, and e:fs TechHub GmbH. Generated with the CARLA simulator, this dataset adopts domain randomization strategies to boost its diversity, and is specifically optimized for 3D object detection tasks using LiDAR point clouds. It contains samples generated across multiple maps, which replicate the configuration of the KITTI dataset, and supports offline usage. Moreover, data processing scripts and final LiDAR variant datasets are provided to facilitate future research and experimental variants.




