AmodalSynthDrive
收藏资源简介:
AmodalSynthDrive是一个专为自动驾驶设计的合成模态感知数据集,由IAV GmbH和弗莱堡大学计算机科学系共同创建。该数据集包含150个驾驶序列,涵盖多种交通、天气和光照条件,提供多视角摄像头图像、3D边界框、激光雷达数据和里程计。数据集支持多种模态场景理解任务,包括新引入的模态深度估计,以增强空间理解。AmodalSynthDrive旨在通过提供丰富的视觉线索和精确的标注,推动模态感知领域的研究,特别是在处理复杂环境和提高自动驾驶系统的安全性和效率方面。
AmodalSynthDrive is a synthetic modal perception dataset specifically designed for autonomous driving, jointly created by IAV GmbH and the Department of Computer Science, University of Freiburg. This dataset includes 150 driving sequences covering diverse traffic scenarios, weather and illumination conditions, and provides multi-view camera images, 3D bounding boxes, LiDAR data and odometry. It supports a range of modal scene understanding tasks, including the newly introduced modal depth estimation to enhance spatial understanding. AmodalSynthDrive aims to advance research in the field of modal perception by providing rich visual cues and precise annotations, particularly in addressing complex environments and improving the safety and efficiency of autonomous driving systems.




