用于测试外观导航方法的新公共数据集
收藏资源简介:
本研究创建了用于测试外观导航方法的新公共数据集,包含三个图像序列。每个序列由多个位置的图像组成,提供了9个图像以及它们之间的相互变换信息,这些变换包括三种横向位移和三种水平相机旋转。图像序列在室外校园环境和室内走廊环境中捕获,室外序列记录了白天和黄昏后的图像,以反映不同的光照条件。该数据集特别为外观导航的横向位移估计任务而设计,弥补了现有数据集的不足。
This study presents a novel public dataset for testing appearance-based navigation methods, which includes three image sequences. Each sequence comprises images captured at multiple locations, providing nine images alongside their mutual transformation information, including three types of lateral displacements and three types of horizontal camera rotations. The image sequences are collected in outdoor campus environments and indoor corridor environments; the outdoor sequences capture images taken during both daytime and post-dusk periods to reflect varying lighting conditions. This dataset is specifically designed for the lateral displacement estimation task in appearance-based navigation, addressing the limitations of existing datasets.




