Motion State
收藏arXiv2024-05-07 更新2024-08-06 收录
下载链接：
http://arxiv.org/abs/2312.17641v2
下载链接
链接失效反馈官方服务：
资源简介：
本研究引入了名为'Motion State'的新基准数据集，专注于多目标跟踪领域。该数据集由西北工业大学自动化学院信息融合重点实验室创建，旨在通过精确的3D坐标信息，评估和提升对象在视频序列中的动态状态识别能力。数据集包含来自KITTI、UAVDT和MOT17等多个场景的视频序列，涵盖了从自动驾驶到无人机监控等多种应用场景。通过这些数据，研究者可以评估和改进多目标跟踪算法在处理静态和动态对象时的性能，特别是在相机移动和对象运动状态复杂的情况下的表现。
This study introduces a novel benchmark dataset named 'Motion State' that focuses on the domain of multi-object tracking. Developed by the Key Laboratory of Information Fusion, School of Automation, Northwestern Polytechnical University, this dataset aims to evaluate and improve the dynamic state recognition ability of objects in video sequences through precise 3D coordinate information. It contains video sequences from multiple scenarios including KITTI, UAVDT and MOT17, covering a wide range of application scenarios from autonomous driving to unmanned aerial vehicle (UAV) monitoring. With this dataset, researchers can assess and refine the performance of multi-object tracking algorithms when handling both static and dynamic objects, especially under complex conditions where the camera is moving and the motion states of objects are intricate.
提供机构：
西北工业大学自动化学院信息融合重点实验室
创建时间：
2023-12-29
