UAVD4L
收藏资源简介:
UAVD4L是由国防科技大学创建的大规模数据集,专为无人机在无GPS环境下的6自由度定位设计。该数据集覆盖约250万平方米的区域,包含多种城市和乡村场景,如建筑物、街道、植被和湖泊。数据集通过高分辨率航拍图像重建纹理3D参考模型,生成合成数据如渲染的RGB和深度图像,以及数字表面模型。数据集的创建过程包括使用DJI M300 RTK无人机和PSDK 102s相机进行图像采集,并通过现代3D重建技术生成模型。UAVD4L旨在解决无人机在无GPS环境下的精确定位问题,适用于救援、监视等多种应用场景。
UAVD4L is a large-scale dataset created by the National University of Defense Technology (NUDT), specifically designed for unmanned aerial vehicle (UAV) 6-degree-of-freedom (6-DoF) positioning in GPS-denied environments. The dataset covers an area of approximately 2.5 million square meters, and includes various urban and rural scenarios such as buildings, streets, vegetation and lakes. It reconstructs textured 3D reference models from high-resolution aerial images, and generates synthetic data including rendered RGB and depth images as well as digital surface models (DSMs). The dataset construction process involves image collection using a DJI M300 RTK UAV and a PSDK 102s camera, followed by model generation with modern 3D reconstruction technologies. UAVD4L aims to address the challenge of precise UAV positioning in GPS-denied environments, and is applicable to multiple application scenarios such as rescue and surveillance.




