A crowdsourced dataset of aerial images with annotated solar photovoltaic arrays and installation metadata

本数据集名为‘A crowdsourced dataset of aerial images with annotated solar photovoltaic arrays and installation metadata’，由巴黎矿业大学 - PSL大学的观测、影响、能源中心 (O.I.E.) 创建。数据集包含超过28000个太阳能光伏安装的航空图像、分割掩码和安装元数据，旨在支持机器学习模型在不同地区和图像源之间准确映射这些安装。数据集内容丰富，包括安装容量、表面、倾斜和方位角等详细信息，适用于构建端到端的太阳能光伏注册表、稳健的安装映射和分析众包数据集。创建过程中，通过两次众包活动收集了多达50个图像标注，以确保标注的准确性。该数据集主要应用于机器学习和众包领域的研究，旨在解决分布偏移问题，促进太阳能光伏安装映射算法的开发。

This dataset is titled "A crowdsourced dataset of aerial images with annotated solar photovoltaic arrays and installation metadata", and was created by the Center for Observations, Impacts and Energy (O.I.E.) of Mines Paris – PSL University. It contains over 28,000 aerial images, segmentation masks and installation metadata for solar photovoltaic (PV) installations, with the goal of supporting machine learning models in accurately mapping these installations across diverse regions and image sources. The dataset features comprehensive detailed attributes including installation capacity, surface area, tilt angle and azimuth angle, and is applicable for building end-to-end solar PV registries, conducting robust installation mapping, and analyzing crowdsourced datasets. During the dataset development process, up to 50 image annotations were collected through two crowdsourcing campaigns to ensure annotation accuracy. This dataset is primarily utilized for research in the fields of machine learning and crowdsourcing, aiming to address distribution shift issues and promote the development of solar PV installation mapping algorithms.

搜集汇总

背景与挑战

背景概述

这是一个由巴黎矿业大学 - PSL大学创建的众包数据集，包含超过28000个太阳能光伏安装的航空图像、分割掩码和详细安装元数据（如容量、表面、倾斜和方位角）。数据集旨在支持机器学习模型准确映射太阳能光伏安装，解决分布偏移问题，适用于构建端到端太阳能光伏注册表和开发映射算法。