DenseUAV

该数据集名为DenseUAV，是首个公开发布的针对无人机自定位任务的数据集。它包含了从14个大学校园收集并标注的超过27,000张无人机视角和卫星视角的图像。该数据集在低空城市环境下对无人机图像进行密集采样，并包含了多种数据增强方法以增强模型训练。规模上，图像总数超过了27,000张，任务专注于无人机自定位。

This dataset, named DenseUAV, is the first publicly released dataset dedicated to the unmanned aerial vehicle (UAV) self-localization task. It contains over 27,000 annotated images captured from both drone and satellite perspectives, collected from 14 university campuses. This dataset performs dense sampling of UAV images in low-altitude urban environments, and incorporates multiple data augmentation methods to enhance model training. In terms of scale, the total number of images exceeds 27,000, and the task focuses on UAV self-localization.

搜集汇总

背景与挑战

背景概述

DenseUAV数据集专注于低空城市环境中的无人机视觉自定位，提供无人机和卫星视角的图像，分为训练、查询和画廊三个子集。该数据集支持多种深度学习方法和评估指标，适用于无人机自定位技术的研究和开发。