Planar Object Tracking Benchmark

本研究提出的Planar Object Tracking Benchmark是一个包含210个视频序列的数据集，用于评估自然环境下的平面物体跟踪算法。该数据集由郑州大学信息工程学院等机构创建，涵盖了30个平面物体，每个物体有7个视频，涉及多种挑战因素如尺度变化、旋转、透视畸变等。数据集的创建过程包括在自然场景中选择物体、拍摄视频并半自动标注地面实况。该数据集主要应用于视觉伺服、视觉SLAM和无人机控制等领域，旨在解决复杂自然环境下平面物体跟踪的问题。

The Planar Object Tracking Benchmark proposed in this research is a dataset consisting of 210 video sequences, which is used to evaluate planar object tracking algorithms in natural environments. Developed by institutions including the School of Information Engineering of Zhengzhou University, this dataset covers 30 planar objects, with 7 video sequences for each object, and involves multiple challenging factors such as scale variation, rotation, perspective distortion, and others. The construction process of the dataset includes selecting objects in natural scenes, shooting videos and performing semi-automatic annotation of ground truth. This dataset is mainly applied in fields such as visual servoing, visual SLAM and unmanned aerial vehicle (UAV) control, aiming to solve the problem of planar object tracking in complex natural environments.