Connecticut Solar PV Semantic Segmentation Dataset

Background Information Annotated overhead imagery dataset for the paper "SolarMapper: estimating solar array location, size, and capacity using deep learning and overhead imagery". This dataset is a subset of the very high resolution aerial imagery provided through the Connecticut Department of Energy and Environmental Protection shared via the University of Connecticut (CT ECO 2016 Imagery & Elevation (uconn.edu)). The original resolution of the imagery is 3 inch (7.62 cm) and we downsampled all of the imagery to a resolution of 30 cm, which is the same resolution as most high resolution satellite imagery. 87 image tiles in total are included in this dataset. We manually annotated all visible solar PV panels with polygons for semantic segmentation purposes. We split the dataset into a training and a validation set at a 2:1 ratio. Details of each subset are shown below: Training set: 57 image tiles, 33.12 km2 of ground area, 608 solar PV arrays, 0.058 km2 of solar PV panel area. Validation set: 30 image tiles, 17.43 km2 of ground area, 1,003 solar PV arrays, 0.091 km2 of solar PV panel area. Additionally, we compared our sampled imagery with Microsoft's US Building Footprints dataset (microsoft/USBuildingFootprints: Computer generated building footprints for the United States (github.com)) to split all image tiles into 3 brackets (high, medium, and low building density) by the number of buildings with in an image tile. More details about how we assigned these image tiles could be found in the paper. Such assignment was only applied to the 30 validation tiles. File Description ct_images_and_labels.zip This .zip file contains all aerial image (.jpg) files and corresponding semantic segmentation annotation mask (.png) files. All .jpg imagery files are 8-bit RGB images and all .png files are binary arrays where 1 is for solar PV pixels and 0 is for non-solar PV pixels. An aerial imagery file and its corresponding mask file share the same filename but have different filename extensions. E.g., 000795_sw.jpg is the aerial imagery file for tile 000795_sw, and 000795_sw.png is the corresponding mask file. tile_assignments.csv This .csv file records if an image tile is in the training or validation set, and if it's a high, medium or low building density tile. There are 3 columns in the .csv file:- tile_name (String): The tile name ID of a tile. This corresponds to filanmes of .jpg aerial imagery files and .png annotation mask files.- training_split (Categorical): Whether the image tile is in the training or the validation set. Values are either 'train' for the training set or "valid" for the validation set.- density_split (Categorical): Which building density bracket does the image tile belong to. Since the building density based assignment was only applied to the 30 validation tiles, all training tiles have "NA" in this column. For the validation tiles, values are either "high", "med", or "low" corresponding to the 3 building density brackets (high, medium and low).背景说明 本数据集为论文《SolarMapper：基于深度学习与航拍影像的光伏阵列位置、规模与容量估算方法》所配套的标注航拍影像数据集。 本数据集源自康涅狄格州能源与环境保护部（Connecticut Department of Energy and Environmental Protection）通过康涅狄格大学公开的超高分辨率航拍影像（CT ECO 2016 Imagery & Elevation，uconn.edu）的部分子集。该原始影像分辨率为3英寸（7.62厘米），我们将所有影像统一下采样至30厘米分辨率，该分辨率与主流高分辨率卫星影像一致。本数据集共包含87个图像瓦片。我们针对语义分割（semantic segmentation）任务，对所有可见的太阳能光伏（Photovoltaic, PV）面板以多边形形式完成人工标注。 我们按照2:1的比例将数据集划分为训练集与验证集，各子集详情如下： 训练集：包含57个图像瓦片，覆盖地面面积33.12平方千米，对应608个太阳能光伏阵列，光伏面板总面积为0.058平方千米。 验证集：包含30个图像瓦片，覆盖地面面积17.43平方千米，对应1003个太阳能光伏阵列，光伏面板总面积为0.091平方千米。 此外，我们结合微软（Microsoft）发布的美国建筑足迹（US Building Footprints）数据集（microsoft/USBuildingFootprints：面向美国的计算机生成建筑足迹数据集，github.com），根据单张图像瓦片内的建筑数量，将所有瓦片划分为高、中、低三类建筑密度区间。关于图像瓦片的具体划分规则，可参阅论文获取更多细节。该密度划分仅应用于30张验证集瓦片。 文件描述 1. ct_images_and_labels.zip 该压缩包包含所有航拍影像文件（.jpg格式）与对应的语义分割标注掩码文件（.png格式）。其中，.jpg影像文件为8位红-绿-蓝（RGB）格式图像；.png掩码文件为二值数组（binary arrays），像素值1代表太阳能光伏面板区域，0代表非光伏区域。航拍影像文件与对应的掩码文件文件名一致，仅扩展名不同。例如，000795_sw.jpg为瓦片000795_sw对应的航拍影像文件，000795_sw.png为其对应的掩码文件。 2. tile_assignments.csv 该逗号分隔值（Comma-Separated Values，CSV）文件记录了各图像瓦片所属的训练/验证集划分，以及其建筑密度区间。文件包含3列： - tile_name（字符串类型）：图像瓦片的名称ID，与对应.jpg航拍影像文件和.png标注掩码文件的文件名一致。 - training_split（分类类型）：标识图像瓦片所属的数据集划分，取值为'train'（训练集）或'valid'（验证集）。 - density_split（分类类型）：标识图像瓦片所属的建筑密度区间。由于密度划分仅应用于30张验证集瓦片，因此所有训练集瓦片的该字段取值为“NA”。对于验证集瓦片，取值分别为“high”（高密度）、“med”（中密度）、“low”（低密度），对应前述三类建筑密度区间。