InCrowd-VI
收藏资源简介:
InCrowd-VI数据集由苏黎世应用科技大学应用信息研究所创建,专门用于评估室内行人密集环境中的SLAM技术,旨在为视觉障碍者的导航提供支持。该数据集包含58个序列,总轨迹长度为5公里,记录时间为1.5小时,涵盖RGB、立体图像和IMU测量数据。数据集捕捉了行人遮挡、人群密度变化、复杂布局和光照变化等现实挑战。通过Meta Aria项目眼镜记录,数据集提供了高精度的地面真实轨迹和半密集的3D点云场景,适用于评估视觉里程计和SLAM算法在复杂室内环境中的性能。
The InCrowd-VI dataset was developed by the Institute of Applied Informatics at Zurich University of Applied Sciences, specifically designed to evaluate SLAM technologies in densely crowded indoor pedestrian environments, aiming to provide navigation support for visually impaired individuals. This dataset includes 58 sequences, with a total trajectory length of 5 kilometers and a total recording duration of 1.5 hours, covering RGB, stereo images, and IMU measurement data. It captures real-world challenges such as pedestrian occlusions, varying crowd densities, complex indoor layouts, and lighting changes. Recorded using Meta Aria Project glasses, the dataset offers high-precision ground-truth trajectories and semi-dense 3D point cloud scenes, which is suitable for assessing the performance of visual odometry and SLAM algorithms in complex indoor environments.




