DROID-W
收藏资源简介:
DROID-W数据集由苏黎世联邦理工学院和微软的研究团队创建,旨在为动态环境下的SLAM研究提供多样化的户外场景数据。该数据集包含通过LiDAR与RGB相机刚性组合采集的真实世界动态场景序列,并额外引入YouTube视频片段以增强'野外'场景评估的多样性。数据采集过程注重捕捉复杂动态对象和杂乱环境,弥补了现有室内基准测试的不足。该数据集主要应用于计算机视觉领域,特别是动态SLAM系统的鲁棒性测试和算法优化,旨在解决传统SLAM在未知动态对象和高度杂乱场景中性能受限的问题。
Developed by a research team from ETH Zurich and Microsoft, the DROID-W dataset is designed to provide diverse outdoor scene data for SLAM research in dynamic environments. It consists of real-world dynamic scene sequences collected via a rigidly integrated LiDAR and RGB camera rig, and additionally incorporates YouTube video clips to boost the diversity of evaluations in "wild" scenarios. The data collection process prioritizes capturing complex dynamic objects and cluttered environments, filling the gaps of existing indoor benchmark datasets. Primarily applied in the field of computer vision, particularly for robustness testing and algorithm optimization of dynamic SLAM systems, this dataset aims to address the performance constraints faced by traditional SLAM systems when dealing with unknown dynamic objects and highly cluttered scenes.

- 1DROID-SLAM in the Wild苏黎世联邦理工学院; 微软 · 2026年



