DELIVER
收藏资源简介:
DELIVER数据集是由香港科技大学(广州)、因萨特大学索菲亚和香港科技大学计算机科学与工程系的研究人员创建的多模态数据集。该数据集包含了深度、激光雷达、多视角、事件和RGB图像,涵盖了五种天气条件(多云、雾天、夜间、雨天和晴天)以及五种角落案例(运动模糊、过曝光、欠曝光、激光雷达抖动和事件低分辨率)。这些数据反映了真实世界传感器性能的挑战。该数据集旨在评估多模态语义分割模型在真实世界自动驾驶场景下的性能和鲁棒性。
The DELIVER Dataset is a multimodal dataset created by researchers from The Hong Kong University of Science and Technology (Guangzhou), Insaat University Sophia, and the Department of Computer Science and Engineering of the Hong Kong University of Science and Technology. It contains depth, LiDAR, multi-view, event, and RGB images, covering five weather conditions (cloudy, foggy, nighttime, rainy, and sunny) as well as five corner cases (motion blur, overexposure, underexposure, LiDAR jitter, and event low resolution). These data reflect the challenges of real-world sensor performance. This dataset is designed to evaluate the performance and robustness of multimodal semantic segmentation models in real-world autonomous driving scenarios.




