ushasi/Remote-Sensing-Datasets

A list of radar and optical satellite datasets for detection, classification, semantic segmentation and instance segmentation tasks.

Remote-Sensing-Datasets

A list of radar and optical satellite datasets for detection, classification, semantic segmentation and instance segmentation tasks.

Satellite imagery datasets according to their corresponding tasks.

A list of radar and optical satellite datasets for computer vision and deep learning. Categories: Instance segmentation, object detection, semantic segmentation, scene classification, retrieval, other.

*The specific datasets could not be accessed. Parts of information on the repository credits are from here, here, and here.

Satellite imagery datasets containing aerial view images.

A list of radar and optical satellite datasets for aerial scene isntance detection, classification, semantic segmentation and instance segmentation tasks.

1. Scene classification

📡 Radar Satellite Datasets :

• FUSAR-Ship Dataset v1.0 - 2020, Hou et al. ↦ Classification

• Statoil/C-CORE Iceberg Classifier Challenge 2 categories ship and iceberg, 2-band HH/HV polarization SAR imagery.

  📡 Optical Satellite Datasets :

• Airbus Wind Turbine Patches
  155k 128x128px image chips with wind turbines (SPOT, 1.5m res.).

• BigEarthNet: Large-Scale Sentinel-2 Benchmark Multilabel (CLC) 2018, 590,326 chips from Sentinel-2 L2A scenes.

• WiDS Datathon 2019 : Detection of Oil Palm Plantations Planet satellite imagery (3m res.)., ca. 20k 256 x 256 px.

• Cactus Aerial Photos 17k aerial photos, 13k cactus, 4k non-actus.

• Functional Map of the World Challenge 63 categories from solar farms to shopping malls, 1 million chips, 4/8 band satellite imagery (0.3m res.).

• EuroSAT 10 land cover categories, 27k 64x64 pixel chips, 3/16 band Sentinel-2 satellite imagery (10m res.).

• AID: Aerial Scene Classification 10000 aerial images within 30 categories collected from Google Earth imagery.

• RESISC45 45 scene categories, 31,500 images (700 per category, 256x256 px), image chips taken from Google Earth.

• Deepsat: SAT-4/SAT-6 airborne datasets 6 land cover categories, 400k 28x28 pixel chips, 4-band RGBNIR aerial imagery (1m res.).

• UC Merced Land Use Dataset 21 land cover categories from agricultural to parkinglot, 100 chips per class, aerial imagery (0.30m res.).

  📡 Multi-sensor Satellite Datasets :

• So2Sat LCZ42 (TUM Munich & DLR, Aug 2018)
  Local climate zone classification, 400k 32x32 pixel chips covering 42 cities (LCZ42 dataset), Sentinel 1 & Sentinel 2 (both 10m res.), 51 GB

• Planet: Understanding the Amazon from Space
  13 land cover categories + 4 cloud condition categories, 4-band (RGB-NIR) satelitte imagery (5m res.).

2. Retrieval

📡 Radar Satellite Datasets :

• coming up!!

📡 Optical Satellite Datasets :

• coming up!!

📡 Multi-sensor Satellite Datasets :

• coming up!!

📡 Cross-sensor Datasets :

• coming up!!

3. Instance Segmentation

🛰: Radar Satellite Datasets:

• HRSID (High-Resolution SAR Images Dataset) - 2020, Wei et al.
• SSDD (SAR Ship Detection Dataset) - 2021, Zhang et al.

📡 Optical Satellite Datasets :

• PASTIS: Panoptic Agricultural Satellite Time Series Sentinel-2 image chip timeseries, panoptic labels (instance index + semantic label for each pixel).
• SpaceNet 7: Multi-Temporal Urban Development Challenge Monthly building footprints and Planet imagery (4m. res) timeseries.
• SpaceNet 4: Off-Nadir Buildings 126k building footprints (Atlanta), 27 WorldView 2 images (0.3m res.).
• SpaceNet 2: Building Detection v2
  685k building footprints, 3/8band Worldview-3 imagery (0.3m res.).
• SpaceNet 1: Building Detection v1 Building footprints (Rio de Janeiro), 3/8band Worldview-3 imagery (0.5m res.).
• RarePlanes: Synthetic Data Takes Flight (CosmiQ Works, A.I.Reverie, June 2020)
  Synthetic (630k planes, 50k images) and real (14.7k planes, 253 Worldview-3 images (0.3m res.), 122 locations, 22 countries) plane annotations & properties and satellite images. Tools. Paper: Shermeyer et al. 2020
• Agriculture-Vision Database & CVPR 2020 challenge 21k aerial farmland images (RGB-NIR, 512x512px chips), label masks for 6 field anomaly patterns..
• iSAID: Large-scale Dataset for Object Detection in Aerial Images 15 categories, 188k instances, object instances and segmentation masks, Google Earth & JL-1 image chips, replaces DOTA dataset.
• xView 2 Building Damage Asessment Challenge (DIUx, Nov 2019) .
  550k building footprints & 4 damage scale categories, 20 global locations and 7 disaster types , Worldview-3 imagery (0.3m res.).
• Microsoft BuildingFootprints Canada & USA & Uganda/Tanzania & Australia (Microsoft, Mar 2019)
  GeoJSON format, delineation based on Bing imagery.
• Airbus Ship Detection Challenge 131k ships, 104k train / 88k test image chips, satellite imagery (1.5m res.).
• Open AI Challenge: Tanzania
  Building footprints & 3 building conditions, RGB UAV imagery - Link to data.
• CrowdAI Mapping Challenge Buildings footprints, RGB satellite imagery.

📡 Multi-sensor Satellite Datasets :

• SpaceNet: Multi-Sensor All-Weather Mapping
  48k building footprints (enhanced 3DBAG dataset, building height attributes), SAR data (4 polarizations) & Worldview-3.
• LPIS agricultural field boundaries Denmark - Netherlands - France
  Denmark: 293 crop/vegetation catgeories, 600k parcels. Netherlands: 294 crop/vegetation catgeories, 780k parcels

4. Object Detection

📡 Radar Satellite Datasets :

• SSDD (SAR Ship Detection Dataset) - 2017, Li et al.
• OpenSARship-1.0, 2.0- 2017, Huang et al.
• SAR-Ship-Dataset - 2019, Wang et al.
• AIR-SARShip -1.0, 2.0 - 2019, Sun et al.
• HRSID (High-Resolution SAR Images Dataset) - 2020, Wei et al.
• LS-SSDD-v1.0 (Large-Scale SAR Ship Detection Dataset) - 2020, Zhang et al.
• SSDD (SAR Ship Detection Dataset) - 2021, Zhang et al.
• DSSDD (Dual-polarimetric SAR Ship Detection Dataset) - 2021, Hu et al.
• SRSDD-v1.0 (SAR Rotation Ship Detection Dataset) - 2021, Lei et al.
• xView3-SAR (Multi-modal SAR Ship Detection + Characterization Dataset) - 2022, Paolo, Lin, Gupta, et. al.

📡 Optical Satellite Datasets :

• Airbus Aircraft Detection 103 images of worlwide airports (Pleiades, 0.5m res., 2560px).
• Airbus Oil Storage Detection Oil storage tank annotations, 98 worldwide images (SPOT, 1.2m res., 2560px).
• AFO - Aerial dataset of floating objects 3647 drone images from 50 scenes, 39991 objects with 6 categories.
• xView 2018 Detection Challenge
  60 categories from helicopter to stadium, 1 million instances, Worldview-3 imagery (0.3m res.).
• Open AI Challenge: Aerial Imagery of South Pacific Islands Tree position & 4 tree species, RGB UAV imagery (0.4m/0.8m res.), multiple AOIs in Tonga
• NOAA Fisheries Steller Sea Lion Population Count 5 sea lion categories, ~ 80k instances, ~ 1k aerial images.
• Stanford Drone Data
  60 aerial UAV videos over Stanford campus and bounding boxes, 6 classes.
• HRSC2016 (High Resolution Ship Collection 2016) - 2016, Liu et al.
• Airbus Ship Detection Challenge Dataset - 2018, Kaggle
• xView Dataset - 2018, Lam et al.
• DOTA (Dataset for Object deTection in Aerial images) - 2018, Xia et al.
• TGRS-HRRSD (High-Resolution Remote Sensing object Detection) - 2018, Kaggle
• MASATI-v2 (MAritime SATellite Imagery dataset) - 2018, Gallego et al.
• DIOR(object Detection In Optical Remote sensing images) - 2019, Li et al.
• FGSD (Fine-Grained Ship Detection) - 2020, Chen et al.
• PSDS (Peruvian Ship Data Set) + MSDS (Mini Ship Data Set) - 2020, Cordova et al.
• ShipRSImageNet - 2021, Zhang et al.
• S2-SHIPS - 2021, Ciocarlan et al.
• GF1-LRSD - 2021, Wu et al.
• VHRShips - 2022, Kizilkaya et al.
• Cars Overhead With Context (COWC) 32k car bounding boxes, aerial imagery (0.15m res.).

📡 Multi-sensor Satellite Datasets :

• xView3 Dark Vessel Detection 2021 Maritime object bounding boxes for 1k Sentinel-1 scenes (VH & VV polarizations), ancillary data (land/ice mask, bathymetry, wind speed, direction, quality).
• NEON Tree Crowns Dataset
  Individual tree crown objects, height&area estimates, 100 million instances, 37 geographic sites across the US, DeepForest, Has LiDAR data.
• NIST DSE Plant Identification with NEON Remote Sensing Data Tree position, tree species and crown parameters, hyperspectral (1m res.) & RGB imagery (0.25m res.), LiDAR point cloud and canopy height model

5. Semantic Segmentation

📡 Radar Satellite Datasets :

• SSDD (SAR Ship Detection Dataset) - 2017, Li et al.

📡 Optical Satellite Datasets :

• FloodNet
  2343 image chips (drone imagery), 10 landcover categories.
• LoveDA
  5987 image chips (Google Earth), 7 landcover categories.
• FloodNet Challenge 2343 UAV images, 2 competition tracks (Binary & semantic flood classification; Object counting & condition recognition)
• Dynamic EarthNet Challenge Weekly Planetscope time-series (3m res.) over 2 years, 75 aois, landcover labels (7 categories), 2 competition tracks (Binary land cover classification & multi-class change detection)
• Sentinel-2 Cloud Mask Catalogue 513 cropped subscenes (1022x1022 pixels) taken randomly from entire 2018 Sentinel-2 archive.
• MiniFrance 2000 very high resolution aerial images over 16 cities in France (50cm res., from IGN BDORTHO).
• LandCoverNet: A Global Land Cover Classification Training Dataset 1980 image chips of 256 x 256 pixels in V1.0 spanning 66 tiles of Sentinel-2.
• LandCover.ai: Dataset for Automatic Mapping of Buildings, Woodlands and Water from Aerial Imagery 41 orthophotos (9000x9000 px) over Poland, Aerial Imagery (25cm & 50cm res.)
• 95-Cloud: A Cloud Segmentation Dataset
  34701 manually segmented 384x384 patches with cloud masks, Landsat 8 imagery (R,G,B,NIR; 30 m res.)
• SkyScapes: Urban infrastructure & lane markings Highly accurate street lane markings & urban infrastructure . Aerial imagery (0.13 m res.).
• Open AI Challenge: Caribbean Predict building roof type (22,553), RGB UAV imagery (4cm res.).
• ALCD Reference Cloud Masks 8 classes (inc. cloud and cloud shadow) for 38 Sentinel-2 scenes (10 m res.).
• Agricultural Crop Cover Classification Challenge 2 main categories corn and soybeans, Landsat 8 imagery (30m res.).
• RoadNet Road network labels, high-res Google Earth imagery, 21 regions.
• SpaceNet 3: Road Network Detection 8000 km of roads in 5 city aois, 3/8band Worldview-3 imagery (0.3m res.).
• DSTL Satellite Imagery Feature Detection Challenge 10 land cover categories from crops to vehicle small, 57 1x1km images, 3/16-band Worldview 3 imagery (0.3m-7.5m res.).
• SPARCS: S2 Cloud Validation data 7 categories, 80 1kx1k px. subset Landsat 8 scenes (30m res.).
• Inria Aerial Image Labeling Building footprint masks, RGB aerial imagery (0.3m res.).

📡 Multi-sensor Satellite Datasets :

• Open Cities AI Challenge
  790k building footprints from Openstreetmap (2 label quality categories), aerial imagery (0.03-0.2m resolution, RGB, 11k 1024x1024 chips, COG format).
• DroneDeploy Segmentation Dataset Drone imagery (0.1m res., RGB), labels & elevation data, baseline model implementation.
• SpaceNet 5: Automated Road Network Extraction & Route Travel Time Estimation 2300 image chips, street geometries with location, shape and estimated travel time, 3/8band Worldview-3 imagery (0.3m res.).
• SEN12MS 180,748 corresponding image triplets containing Sentinel-1 (VV&VH), Sentinel-2 (all bands, cloud-free), and MODIS-derived land cover maps (IGBP, LCCS, 17 classes, 500m res.).
• Slovenia Land Cover Classification 10 land cover classes, temporal stack of hyperspectral Sentinel-2 imagery.
• Urban 3D Challenge
  157k building footprint masks, RGB orthophotos (0.5m res.), DSM/DTM.
• ISPRS Potsdam 2D Semantic Labeling Contest 6 urban land cover classes, raster mask labels, 4-band RGB-IR aerial imagery (0.05m res.) & DSM, 38 image patches

6. Other Focus / Multiple Tasks

• SEN12MS-CR & SEN12MS-CR-TS
  A multi-modal and mono-temporal data set for cloud removal. Sentinel-1 & Sentinel-2, 2018. 175 globally distributed aois.
• IEEE Data Fusion Contest 2022 Semi-supervised semantic segmentation, 19 cities and surroundings with multi-sensor tiles (VHR Aerial imagery 50cm res., Elevation model) & per pixel labels (contains landcover / landuse classes from UrbanAtlas 2012),
• IEEE Data Fusion Contest 2021 Detection of settlements without electricity, 98 multi-temporal/multi-sensor tiles ( Sentinel-1, Sentinel-2, Landsat-8, VIIRS), per chip & per pixel labels (contains buildings, presence electricity).
• University-1652: Drone-based Geolocalization (Image Retrieval)
  Corresponding imagery from drone, satellite and ground camera of 1,652 university buildings.
• IEEE Data Fusion Contest 2020 Land cover classification based on SEN12MS dataset (see category Semantic Segmentation on this list), low- and high-resolution tracks.
• IEEE Data Fusion Contest 2019 Multiple tracks: Semantic 3D reconstruction, Semantic Stereo, 3D-Point Cloud Classification. Worldview-3 (8-band, 0.35cm res.) satellite imagery, LiDAR (0.80m pulse spacing, ASCII format), semantic labels, urban setting USA, baseline methods provided,
• IEEE Data Fusion Contest 2018 20 land cover categories by fusing three data sources: Multispectral LiDAR, Hyperspectral (1m), RGB imagery (0.05m res.).
• DEEPGLOBE - 2018 Satellite Challange Three challenge tracks: Road Extraction, Building Detection, Land cover classification.
• TiSeLaC: Time Series Land Cover Classification Challenge Land cover time series classification (9 categories), Landsat-8 (23 images time series, 10 band features, 30m res.).
• Multi-View Stereo 3D Mapping Challenge Develop a Multi-View Stereo (MVS) 3D mapping algorithm that can convert high-resolution Worldview-3 satellite images to 3D point clouds, 0.2m lidar ground truth data.
• Draper Satellite Image Chronology Predict the chronological order of images taken at the same locations over 5 days, Kaggle kernels

Go To TOP