KITTI InstanceMotSeg Dataset
收藏资源简介:
KITTI InstanceMotSeg数据集是由Zewail City of Science and Technology创建的,专注于自动驾驶中的单目实例运动分割任务。该数据集包含12,919个样本,提供了多类别的实例级运动分割标注,旨在通过运动线索实现对未知物体的检测。数据集的创建过程涉及对3D世界坐标系统中运动的精确估计,以提高标注的准确性。该数据集主要应用于自动驾驶领域,用于解决复杂场景下的运动物体分割问题,提高轨迹规划的准确性和安全性。
The KITTI InstanceMotSeg dataset, developed by Zewail City of Science and Technology, focuses on the monocular instance motion segmentation task in autonomous driving. This dataset consists of 12,919 samples and provides multi-category instance-level motion segmentation annotations, aiming to detect unknown objects via motion cues. The construction of this dataset involves accurate estimation of motion in the 3D world coordinate system to improve the accuracy of annotations. This dataset is primarily applied in the autonomous driving domain, to address the issue of moving object segmentation in complex scenarios and enhance the accuracy and safety of trajectory planning.




