CRUW3D

CRUW3D数据集是由华盛顿大学等机构联合创建的，包含66,000个同步且校准良好的摄像头、雷达和激光雷达帧，适用于自动驾驶中的3D目标感知任务。该数据集不仅包含3D位置信息，还包含时空语义信息，支持机器学习模型在融合摄像头和雷达信息后生成更可靠的目标感知结果。数据集涵盖多种驾驶场景和光照条件，旨在解决自动驾驶中的可靠性和鲁棒性问题，特别是在不同天气和光照条件下的感知挑战。

The CRUW3D dataset was jointly created by the University of Washington and other institutions. It contains 66,000 synchronized and calibrated camera, radar, and LiDAR frames, and is tailored for 3D object perception tasks in autonomous driving. Besides 3D position information, the dataset also includes spatiotemporal semantic information, enabling machine learning models to generate more reliable object perception results by fusing camera and radar data. The dataset covers diverse driving scenarios and lighting conditions, aiming to address the reliability and robustness issues in autonomous driving, especially the perception challenges under varying weather and lighting conditions.