Awesome Satellite Imagery Datasets
收藏资源简介:
列表包含航空和卫星图像数据集,带有计算机视觉和深度学习所需的标注。数据集按类别(实例分割、目标检测、语义分割、场景分类、其他)最新数据集排在顶部。
The list comprises datasets of aerial and satellite imagery, annotated for computer vision and deep learning applications. Datasets are categorized (instance segmentation, object detection, semantic segmentation, scene classification, others) with the most recent datasets listed at the top.
数据集概述
1. 实例分割
-
FloodNet
数据来源:University of Maryland
发布时间:Jun 2021
数据内容:2343 image chips (drone imagery), 10 landcover categories
相关论文:Rahnemoonfar et al., 2021 -
PASTIS : Panoptic Agricultural Satellite TIme Series
数据来源:IGN
发布时间:Jul 2021
数据内容:124,422 Agricultural parcels, 2,433 Sentinel-2 image chip timeseries, France, panoptic labels
相关论文:Garnot & Landrieu 2021 -
xView3 Dark Vessel Detection 2021
数据来源:xView3 Team
发布时间:Aug 2021
数据内容:Maritime object bounding boxes for 1k Sentinel-1 scenes
2. 目标检测
-
Airbus Aircraft Detection
数据来源:Airbus
发布时间:Mar 2021
数据内容:Aircraft bounding boxes, 103 images of worldwide airports -
Airbus Oil Storage Detection
数据来源:Airbus
发布时间:Mar 2021
数据内容:Oil storage tank annotations, 98 worldwide images -
xView3 Dark Vessel Detection 2021
数据来源:xView3 Team
发布时间:Aug 2021
数据内容:Maritime object bounding boxes for 1k Sentinel-1 scenes
3. 语义分割
-
FloodNet
数据来源:University of Maryland
发布时间:Jun 2021
数据内容:2343 image chips (drone imagery), 10 landcover categories
相关论文:Rahnemoonfar et al., 2021 -
LoveDA
数据来源:Wuhan University
发布时间:Oct 2021
数据内容:5987 image chips (Google Earth), 7 landcover categories
相关论文:Wang et al., 2021 -
FloodNet Challenge
数据来源:UMBC, Microsoft, Texas A&M, Dewberry
发布时间:May 2021
数据内容:2343 UAV images from after Hurricane Harvey, landcover labels -
Dynamic EarthNet Challenge
数据来源:Planet, DLR, TUM
发布时间:April 2021
数据内容:Weekly Planetscope time-series over 2 years, 75 aois, landcover labels -
Sentinel-2 Cloud Mask Catalogue
数据来源:Francis, A., et al.
发布时间:Nov 2020
数据内容:513 cropped subscenes (1022x1022 pixels) taken randomly from entire 2018 Sentinel-2 archive -
MiniFrance
数据来源:Université Bretagne-Sud and ONERA
发布时间:Jul 2020
数据内容:2000 very high resolution aerial images over 16 cities in France
相关论文:Castillo-Navarro et al., 2021 -
LandCoverNet: A Global Land Cover Classification Training Dataset
数据来源:Alemohammad S.H., et al.
发布时间:Jul 2020
数据内容:1980 image chips of 256 x 256 pixels in V1.0 spanning 66 tiles of Sentinel-2
相关论文:Alemohammad S.H., et al., 2020 -
LandCover.ai: Dataset for Automatic Mapping of Buildings, Woodlands and Water from Aerial Imagery
数据来源:Boguszewski, A., et al.
发布时间:May 2020
数据内容:41 orthophotos (9000x9000 px) over Poland, Aerial Imagery
相关论文:Boguszewski et al., 2020 -
95-Cloud: A Cloud Segmentation Dataset
数据来源:S. Mohajerani et. all
发布时间:Jan 2020
数据内容:34701 manually segmented 384x384 patches with cloud masks
相关论文:Mohajerani et al. 2021 -
Open Cities AI Challenge
数据来源:GFDRR
发布时间:Mar 2020
数据内容:790k building footprints from Openstreetmap, aerial imagery -
DroneDeploy Segmentation Dataset
数据来源:DroneDeploy
发布时间:Dec 2019
数据内容:Drone imagery, labels (7 land cover catageories) & elevation data -
SkyScapes: Urban infrastructure & lane markings
数据来源:DLR
发布时间:Nov 2019
数据内容:Highly accurate street lane markings & urban infrastructure
相关论文:Azimi et al. 2019 -
Open AI Challenge: Caribbean
数据来源:MathWorks, WeRobotics, Wordlbank, DrivenData
发布时间:Dec 2019
数据内容:Predict building roof type of provided building footprints, RGB UAV imagery -
SpaceNet 5: Automated Road Network Extraction & Route Travel Time Estimation
数据来源:CosmiQ Works, Maxar, Intel, AWS
发布时间:Sep 2019
数据内容:2300 image chips, street geometries with location, shape and estimated travel time -
SEN12MS
数据来源:TUM
发布时间:Jun 2019
数据内容:180,748 corresponding image triplets containing Sentinel-1, Sentinel-2, and MODIS-derived land cover maps
相关论文:Schmitt et al. 2018 -
Slovenia Land Cover Classification
数据来源:Sinergise
发布时间:Feb 2019
数据内容:10 land cover classes, temporal stack of hyperspectral Sentinel-2 imagery -
ALCD Reference Cloud Masks
数据来源:CNES
发布时间:Oct 2018
数据内容:8 classes for 38 Sentinel-2 scenes
相关论文:Baetens et al. 2019 -
Agricultural Crop Cover Classification Challenge
数据来源:CrowdANALYTIX
发布时间:Jul 2018
数据内容:2 main categories corn and soybeans, Landsat 8 imagery -
RoadNet
数据来源:Wuhan
发布时间:Oct 2018
数据内容:Road network labels, high-res Google Earth imagery
相关论文:Liu et al. 2018 -
SpaceNet 3: Road Network Detection
数据来源:CosmiQ Works, Radiant Solutions
发布时间:Feb 2018
数据内容:8000 km of roads in 5 city aois
相关论文:Van Etten et al. 2018 -
Urban 3D Challenge
数据来源:USSOCOM
发布时间:Dec 2017
数据内容:157k building footprint masks, RGB orthophotos -
DSTL Satellite Imagery Feature Detection Challenge
数据来源:Dstl
发布时间:Feb 2017
数据内容:10 land cover categories from crops to vehicle small -
SPARCS: S2 Cloud Validation data
数据来源:USGS
发布时间:2016
数据内容:7 categories, 80 1kx1k px. subset Landsat 8 scenes
相关论文:Hughes, J.M. & Hayes D.J. 2014 -
Biome: L8 Cloud Cover Validation data
数据来源:USGS
发布时间:2016
数据内容:4 cloud categories, 96 Landsat 8 scenes
相关论文:Foga et al. 2017 -
Inria Aerial Image Labeling
数据来源:inria.fr
数据内容:Building footprint masks, RGB aerial imagery -
ISPRS Potsdam 2D Semantic Labeling Contest
数据来源:ISPRS
数据内容:6 urban land cover classes, raster mask labels, 4-band RGB-IR aerial imagery & DSM




