视觉无人机自主巡检与裂缝识别数据

面向基于视觉的机器人环境建模与定位导航基础研究，为突破复杂环境下机器人视觉实时高精度定位技术瓶颈，基于精灵 PHANTOM 4 RTK多旋翼无人机进行数据采集，数据集覆盖场景是同济大学四平路校区中心区域的一座五层办公楼（该建筑周边被近似高度的建筑和树木近距离包围，相邻对象之间形成狭窄的巷道，并在一定高度范围相互遮挡，不同区域有明显的高度差异、结构丰富、部件多样、纹理多变）。该数据集包含96幅相片，数据量为986 M，主要用于视觉机器人的自主定位导航与目标识别。

This dataset is designed for fundamental research on vision-based robotic environmental modeling, localization and navigation. To break through the technical bottleneck of real-time high-precision visual localization for robots in complex environments, data was collected using a DJI PHANTOM 4 RTK multi-rotor unmanned aerial vehicle. The dataset covers a five-story office building located in the central area of the Siping Road Campus of Tongji University. The building is surrounded by buildings and trees of similar height in close proximity, forming narrow alleyways between adjacent objects, with mutual occlusion within a certain height range. Significant height differences exist across different areas of the scene, which features rich structures, diverse components and variable textures. This dataset contains 96 images with a total data size of 986 MB, and is primarily utilized for autonomous localization, navigation and target recognition of vision-based robots.

搜集汇总

背景与挑战

背景概述

该数据集专注于基于视觉的无人机自主巡检与裂缝识别，旨在支持机器人环境建模与定位导航研究。数据集包含96幅相片，总数据量为986MB，采集自同济大学四平路校区的一个复杂办公楼场景，该场景具有结构丰富、纹理多变的特点，适用于视觉机器人的高精度定位和目标识别任务。数据以jpg格式提供，通过观测监测方式生产，主要服务于摄影测量与遥感技术领域的应用。