EVIMO2
收藏资源简介:
EVIMO2是由马里兰大学高级计算机研究所开发的室内场景事件相机数据集,包含41分钟的数据,由三个640×480事件相机和一个2080×1552经典彩色相机记录。数据集包含173个序列,分为三个类别:独立移动的家用物品、静态场景和浅层场景中的基本运动。数据集的创建结合了3D扫描和Vicon运动捕捉系统的高精度跟踪,提供了详细的深度图和光流场。EVIMO2特别适用于研究运动和物体分割、光流、结构从运动以及视觉惯性里程计,旨在解决室内动态场景中的复杂视觉问题。
EVIMO2 is an indoor scene event camera dataset developed by the Institute for Advanced Computer Studies at the University of Maryland. It contains 41 minutes of data, recorded by three 640×480 event cameras and one 2080×1552 conventional color camera. The dataset includes 173 sequences divided into three categories: independently moving household objects, static scenes, and basic motions in shallow scenes. The dataset was constructed using 3D scanning and high-precision tracking from the Vicon motion capture system, providing detailed depth maps and optical flow fields. EVIMO2 is particularly suitable for research on motion and object segmentation, optical flow, structure from motion, and visual-inertial odometry, aiming to solve complex visual problems in indoor dynamic scenes.




