前言数据是人工智能时代的石油，随着汽车行业的发展、自动驾驶商业场景的落地，自动驾驶算法变得尤为重要，想要打磨自动驾驶算法，就需要大量的场景数据支撑。作者对过去使用、积累的自动驾驶开源数据集做了总结，以下45个自动驾驶开源数据集供大家参考。4Seasons题目：4Seasons: A Cross-Season Dataset for Multi-Weather SLAM in Autonomous Driving名称：4Seasons：自动驾驶中多天气 SLAM 的跨季节数据集论文：arxiv.org/abs/2009.0636主页：4seasons-dataset.com/
4SeasonsA2D2题目：A2D2: Audi Autonomous Driving Dataset名称：A2D2：奥迪自动驾驶数据集论文：arxiv.org/abs/2004.0632主页：a2d2.audi/a2d2/en.html下载：a2d2.audi/a2d2/en/downl
A2D2A3D题目：A3D Dataset: Towards Autonomous Driving in Challenging Environments名称：A*3D 数据集：在具有挑战性的环境中实现自动驾驶论文：arxiv.org/abs/1909.0754主页：gas.graviti.cn/dataset/
A*3DApolloScape题目：The ApolloScape Open Dataset for Autonomous Driving and its Application名称：用于自动驾驶的 ApolloScape 开放数据集及其应用论文：arxiv.org/abs/1803.0618官方：http://apolloscape.auto下载：apolloscape.auto/tracki
ApolloScapeBDD100K题目：BDD100K: A Diverse Driving Dataset for Heterogeneous Multitask Learning名称：BDD100K：异构多任务学习的多样化驱动数据集论文：arxiv.org/abs/1805.0468主页：bdd-data.berkeley.edu/下载：bdd-data.berkeley.edu/p
BDD100KBLVD题目：BLVD: Building A Large-scale 5D Semantics Benchmark for Autonomous Driving名称：BLVD：为自动驾驶构建大规模 5D 语义基准论文：arxiv.org/abs/1903.0640主页：github.com/VCCIV/BLVD
BLVDBoreas题目：Boreas: A Multi-Season Autonomous Driving Dataset名称：Boreas：多季节自动驾驶数据集论文：arxiv.org/abs/2203.1016主页：boreas.utias.utoronto.ca
BoreasCADC题目：Canadian Adverse Driving Conditions Dataset名称：加拿大不良驾驶条件数据集论文：arxiv.org/abs/2001.1011主页：scale.com/open-datasets
CADCCityScapes题目：The Cityscapes Dataset for Semantic Urban Scene Understanding名称：用于语义城市场景理解的 Cityscapes 数据集论文：arxiv.org/abs/1604.0168主页：cityscapes-dataset.com/下载：cityscapes-dataset.com/
CityScapesCODA题目：CODA: A Real-World Road Corner Case Dataset for Object Detection in Autonomous Driving名称：CODA：用于自动驾驶目标检测的真实道路拐角案例数据集论文：arxiv.org/abs/2203.0772主页：coda-dataset.github.io/
CODAComma2K19题目：A Commute in Data: The comma2k19 Dataset名称：数据通勤：comma2k19 数据集论文：arxiv.org/abs/1812.0575代码：github.com/commaai/comm主页：comma.ai/
Comma2K19DGL-MOTS题目：DG-Labeler and DGL-MOTS Dataset: Boost the Autonomous Driving Perception名称：DG-Labeler 和 DGL-MOTS 数据集：提升自动驾驶感知论文：arxiv.org/abs/2110.0779
DGL-MOTSFordMultiAV题目：Ford Multi-AV Seasonal Dataset名称：福特多 AV 季节性数据集论文：arxiv.org/abs/2003.0796主页：avdata.ford.com/
FordMultiAVH3D题目：The H3D Dataset for Full-Surround 3D Multi-Object Detection and Tracking in Crowded Urban Scenes名称：用于拥挤城市场景中的全环绕 3D 多目标检测和跟踪的 H3D 数据集论文：arxiv.org/abs/1903.0156
H3DIDDA题目：IDDA: a large-scale multi-domain dataset for autonomous driving名称：IDDA：用于自动驾驶的大规模多域数据集论文：arxiv.org/abs/2004.0829主页：idda-dataset.github.io/
IDDAKITTI题目：Vision meets Robotics: The KITTI Dataset论文：视觉与机器人技术：KITTI数据集论文：cvlibs.net/publications下载：cvlibs.net/datasets/kit
KITTIKITTI-360题目：KITTI-360: A Novel Dataset and Benchmarks for Urban Scene Understanding in 2D and 3D名称：KITTI-360：2D 和 3D 城市场景理解的新数据集和基准论文：arxiv.org/abs/2109.1341主页：cvlibs.net/datasets/kit
KITTI-360KITTI-InstanceMotSeg题目：Monocular Instance Motion Segmentation for Autonomous Driving: KITTI InstanceMotSeg Dataset and Multi-task Baseline名称：用于自动驾驶的单目实例运动分割：KITTI InstanceMotSeg 数据集和多任务基线论文：arxiv.org/abs/2008.0700主页：sites.google.com/view/i
KITTI-InstanceMotSegKITTI-MOTS题目：MOTS: Multi-Object Tracking and Segmentation名称：MOTS：多对象跟踪和分割论文：arxiv.org/abs/1902.0360主页：cvlibs.net/datasets/kit项目：vision.rwth-aachen.de/p
KITTI-MOTSKITTI-SceneFlow题目：Object Scene Flow for Autonomous Vehicles名称：自动驾驶车辆的目标场景流论文：cvlibs.net/publications主页：cvlibs.net/datasets/kit
KITTI-SceneFlowLIBRE题目：LIBRE: The Multiple 3D LiDAR Dataset名称：LIBRE：多 3D LiDAR 数据集论文：arxiv.org/abs/2003.0612
LIBRELyftL5题目：One Thousand and One Hours: Self-driving Motion Prediction Dataset名称：一千零一小时：自动驾驶运动预测数据集论文：arxiv.org/abs/2006.1448主页：level-5.global/
LyftL5NEOLIX题目：The NEOLIX Open Dataset for Autonomous Driving名称：NEOLIX 自动驾驶开放数据集论文：arxiv.org/abs/2011.1352
NEOLIXNuScenes题目：nuScenes: A multimodal dataset for autonomous driving名称：nuScenes：用于自动驾驶的多模式数据集论文：arxiv.org/abs/1903.1102主页：nuscenes.org/
NuScenesONCE题目：One Million Scenes for Autonomous Driving: ONCE Dataset名称：一百万个自动驾驶场景：ONCE 数据集论文：arxiv.org/abs/2106.1103主页：once-for-auto-driving.github.io
ONCEOpenLane题目：PersFormer: 3D Lane Detection via Perspective Transformer and the OpenLane Benchmark名称：PersFormer：通过 Perspective Transformer 和 OpenLane 基准进行 3D 车道检测论文：arxiv.org/abs/2203.1108项目：github.com/OpenPercepti代码：github.com/OpenPercepti
OpenLaneOPV2V题目：OPV2V: An Open Benchmark Dataset and Fusion Pipeline for Perception with Vehicle-to-Vehicle Communication名称：OPV2V：用于车对车通信感知的开放基准数据集和融合管道论文：arxiv.org/abs/2109.0764项目：github.com/DerrickXuNu/主页：mobility-lab.seas.ucla.edu
OPV2VOxford-RobotCar题目：The Oxford Radar RobotCar Dataset: A Radar Extension to the Oxford RobotCar Dataset名称：Oxford Radar RobotCar 数据集：牛津 RobotCar 数据集的雷达扩展论文：arxiv.org/abs/1909.0130主页：robotcar-dataset.robots.ox.ac.uk下载：oxford-robotics-institute.github.io主页：oxford-robotics-institute.github.io
Oxford-RobotCarPandaSet题目：PandaSet: Advanced Sensor Suite Dataset for Autonomous Driving名称：PandaSet：用于自动驾驶的高级传感器套件数据集论文：arxiv.org/abs/2112.1261主页：pandaset.org/主页：scale.com/open-datasets
PandaSetPASCAL3D+题目：Beyond PASCAL: A Benchmark for 3D Object Detection in the Wild名称：Beyond PASCAL：野外3D物体检测的基准论文：cvgl.stanford.edu/paper论文：ieeexplore.ieee.org/doc
PASCAL3D+PixSet题目：PixSet : An Opportunity for 3D Computer Vision to Go Beyond Point Clouds With a Full-Waveform LiDAR Dataset名称：PixSet：3D 计算机视觉利用全波形 LiDAR 数据集超越点云的机会论文：arxiv.org/abs/2102.1201主页：https://leddartech.com/soluti...
PixSetRADIATE题目：RADIATE: A Radar Dataset for Automotive Perception in Bad Weather名称：辐射：恶劣天气下汽车感知的雷达数据集论文：arxiv.org/abs/2010.0907主页：pro.hw.ac.uk/radiate/
RADIATEROAD题目：ROAD: The ROad event Awareness Dataset for Autonomous Driving名称：ROAD：自动驾驶的 ROad 事件感知数据集论文：arxiv.org/abs/2102.1158主页：github.com/gurkirt/3D-R
ROADRope3D题目：Rope3D: TheRoadside Perception Dataset for Autonomous Driving and Monocular 3D Object Detection Task名称：Rope3D：用于自动驾驶和单目 3D 目标检测任务的路边感知数据集论文：arxiv.org/abs/2203.1360主页：thudair.baai.ac.cn/rope
Rope3DSemanticKITTI题目：SemanticKITTI: A Dataset for Semantic Scene Understanding of LiDAR Sequences名称：SemanticKITTI：用于对 LiDAR 序列进行语义场景理解的数据集论文：arxiv.org/abs/1904.0141主页：competitions.codalab.org主页：semantic-kitti.org/
SemanticKITTISODA10M题目：SODA10M: A Large-Scale 2D Self/Semi-Supervised Object Detection Dataset for Autonomous Driving名称：SODA10M：用于自动驾驶的大规模二维自/半监督目标检测数据集论文：arxiv.org/abs/2106.1111主页：soda-2d.github.io/
SODA10MSynWoodScape题目：SynWoodScape: Synthetic Surround-view Fisheye Camera Dataset for Autonomous Driving名称：SynWoodScape：用于自动驾驶的合成环视鱼眼相机数据集论文：arxiv.org/abs/2203.0505
SynWoodScapeTJ4DRadSet题目：TJ4DRadSet: A 4D Radar Dataset for Autonomous Driving名称：TJ4DRadSet：用于自动驾驶的 4D 雷达数据集论文：arxiv.org/abs/2204.1348
TJ4DRadSetTopoBoundary题目：Topo-boundary: A Benchmark Dataset on Topological Road-boundary Detection Using Aerial Images for Autonomous Driving名称：地形边界：使用航空图像进行自动驾驶拓扑道路边界检测的基准数据集论文：arxiv.org/abs/2103.1711
TopoBoundaryUtbmRobocar题目：EU Long-term Dataset with Multiple Sensors for Autonomous Driving名称：用于自动驾驶的具有多个传感器的欧盟长期数据集论文：arxiv.org/abs/1909.0333主页：epan-utbm.github.io/utb
UtbmRobocarV2X-Sim题目：V2X-Sim: A Virtual Collaborative Perception Dataset for Autonomous Driving名称：V2X-Sim：用于自动驾驶的虚拟协作感知数据集论文：arxiv.org/abs/2202.0844
V2X-SimVirtual-KITTI-2题目：Virtual KITTI 2名称：虚拟 KITTI 2论文：arxiv.org/abs/2001.1077主页：paperswithcode.com/data
Virtual-KITTI-2Waymo题目：Scalability in Perception for Autonomous Driving: Waymo Open Dataset名称：自动驾驶感知的可扩展性：Waymo 开放数据集论文：arxiv.org/abs/1912.0483主页：waymo.com/open下载：waymo.com/open/download
WaymoWaymoMotion题目：Large Scale Interactive Motion Forecasting for Autonomous Driving : The Waymo Open Motion Dataset名称：自动驾驶的大规模交互式运动预测：Waymo Open Motion 数据集论文：arxiv.org/abs/2104.1013
WaymoMotionWoodScape题目：WoodScape: A multi-task, multi-camera fisheye dataset for autonomous driving名称：WoodScape：用于自动驾驶的多任务、多摄像头鱼眼数据集论文：arxiv.org/abs/1905.0148主页：gas.graviti.cn/dataset/
WoodScapeDATATANG高质量数据是车路协同自动驾驶技术的关键，能够持续优化车路协同的算法，助力自动驾驶系统迭代升级。
数据采集是数据处理的核心源头，代表着数据全生命周期的合法合规度。数据堂获得国家测绘资质，配备激光雷达数据自采车辆，可以帮助企业采集高质量的自动驾驶道路数据。
相关文章
