大规模自动驾驶数据集

<p>该数据集包含在波士顿和新加坡采集的 1000 个场景的 3D 边界框数据。每个场景长20秒，注释为2Hz。总共28130个用于训练的样本，6019个用于验证的样本和6008个用于测试的样本。该数据集具有完整的自动驾驶车辆数据套件: 32光束激光雷达，6个摄像头和具有完整360 ° 覆盖的雷达。3D对象检测挑战评估10个类别的性能: 汽车，卡车，公共汽车，拖车，建筑车辆，行人，摩托车，自行车，交通锥和障碍物。nuScenes-大规模自动驾驶数据集</p><p>数据来源：公开网站<a href="https://opendatalab.org.cn/OpenDataLab/nuScenes" rel="noopener noreferrer" target="_blank">https://opendatalab.org.cn/OpenDataLab/nuScenes</a></p><p>数据规模：约1000个场景</p><p>数据特点：该数据集具有完整的自动驾驶车辆数据套件，3D对象检测挑战评估多类别的性能</p><p>应用场景：自动驾驶汽车训练</p>

This is the nuScenes large-scale autonomous driving dataset, which houses 3D bounding box data from 1000 scenarios collected in Boston and Singapore. Each scenario spans 20 seconds and is annotated at a rate of 2 Hz. In total, the dataset provides 28130 training samples, 6019 validation samples, and 6008 test samples. It features a full autonomous driving sensor suite: 32-beam LiDAR, 6 cameras, and radar with full 360° coverage. The accompanying 3D object detection challenge evaluates model performance across 10 categories: car, truck, bus, trailer, construction vehicle, pedestrian, motorcycle, bicycle, traffic cone, and obstacle.
Data Source: Public website <a href="https://opendatalab.org.cn/OpenDataLab/nuScenes" rel="noopener noreferrer" target="_blank">https://opendatalab.org.cn/OpenDataLab/nuScenes</a>
Data Scale: Approximately 1000 scenarios
Data Characteristics: This dataset is equipped with a complete autonomous driving vehicle sensor suite, and its 3D object detection challenge assesses performance across multiple object categories
Application Scenarios: Autonomous vehicle training