无人机遥感影像 影像分割数据集

DroneDeploy无人机遥感影像分割数据集由DroneDeploy于2019年发布，包含140条高分辨率航拍数据（10厘米/像素），涵盖RGB图像、高程数据和语义标签。数据分为小型（0.5GB）和中型（9.0GB）两个版本，按train/val/test划分，适用于快速测试与训练。标签包含BUILDING、VEGETATION、WATER等6类地物，以颜色编码存储于PNG文件，需转换为类别索引。原始图像需切分为300x300像素块并可进行翻转、旋转增强。该数据集广泛应用于城市规划、农业监测、建筑检测、环境与基础设施管理等场景。The DroneDeploy UAV Remote Sensing Image Segmentation Dataset was published by DroneDeploy in 2019. It contains 140 high-resolution aerial imagery samples with a spatial resolution of 10 cm/pixel, covering RGB images, elevation data, and semantic labels. The dataset is offered in two variants: a small version (0.5 GB) and a medium version (9.0 GB), which are split into training, validation, and test subsets for rapid model testing and training. The semantic labels include 6 categories of ground features such as BUILDING, VEGETATION, and WATER, which are stored in color-coded PNG files and require conversion into category indices. The original images should be cropped into 300×300 pixel patches, and data augmentation operations including flipping and rotation can be applied. This dataset is widely used in scenarios such as urban planning, agricultural monitoring, building detection, environmental management, and infrastructure administration.

搜集汇总

背景与挑战

背景概述

该数据集是DroneDeploy于2019年发布的无人机遥感影像分割数据集，包含140条高分辨率航拍数据（10厘米/像素），涵盖RGB图像、高程数据和语义标签，标签分为BUILDING、VEGETATION、WATER等6类地物。数据分为小型（0.5GB）和中型（9.0GB）两个版本，已按train/val/test划分，适用于快速测试与训练，广泛应用于城市规划、农业监测和建筑检测等场景。